Download A l - EN / Bilkent University

MODELING RF HEATING OF ACTIVE
IMPLANTABLE MEDICAL DEVICES
DURING MRI USING SAFETY INDEX
A THESIS
SUBMITTED TO THE DEPARTMENT OF ELECTRICAL AND
ELECTRONICS ENGINEERING
AND THE INSTITUTE OF ENGINEERING AND SCIENCES
OF BILKENT UNIVERSITY
IN PARTIAL FULLFILMENT OF THE REQUIREMENTS
FOR THE DEGREE OF
MASTER OF SCIENCE
By
Halise Irak
August 2007
I certify that I have read this thesis and that in my opinion it is fully adequate, in
scope and in quality, as a thesis for the degree of Master of Science.
Prof. Dr. Ergin Atalar (Supervisor)
I certify that I have read this thesis and that in my opinion it is fully adequate, in
scope and in quality, as a thesis for the degree of Master of Science.
Prof. Dr. Nevzat Gençer
I certify that I have read this thesis and that in my opinion it is fully adequate, in
scope and in quality, as a thesis for the degree of Master of Science.
Assist. Prof. Dr. Vakur Ertürk
Approved for the Institute of Engineering and Sciences:
Prof. Dr. Mehmet B. Baray
Director of Institute of Engineering and Sciences
ii
ABSTRACT
MODELING RF HEATING OF ACTIVE IMPLANTABLE
MEDICAL DEVICES DURING MRI
USING SAFETY INDEX
Halise Irak
M.S. in Electrical and Electronics Engineering
Supervisor: Prof. Dr. Ergin Atalar
August 2007
Magnetic Resonance Imaging (MRI) is known as a safe imaging
modality that can be hazardous for patients with active implantable medical
devices, such as a pacemakers or deep brain stimulators. The primary reason for
that is the radio frequency (RF) heating at the tips of the implant leads. In the
past, this problem has been analyzed with phantom, animal and human
experiments. The amount of temperature rise at the lead tip of these implants,
however, has not been theoretically analyzed. In this thesis, a simple
approximate formula for the safety index of implants, which is the temperature
increase at the implant lead tip per unit deposited power in the tissue without the
implant in place, was derived.
For that purpose, an analytical quadrature birdcage coil model was
developed and the longitudinal incident electric field distribution inside the body
was formularized as follows:
E z ( R) = − ω µ H − R
in which ω is the angular frequency, µ is the magnetic permeability of the tissue,
H- is the left hand rotating component of the RF magnetic field and R is the
radial distance from the center of the body. This formula was examined by
simulations and phantom experiments. The analytical, simulation and
experimental results of that model are in good agreement.
iii
Then, depending on the quadrature birdcage coil model safety index (SI)
formula for active implants with short leads was derived as shown below:
SI =
∆Tmax
1
=
Rl + Ae jθ
SARpeak 2α ct Rb2
2
f ( Dv)
where ∆Tmax is the maximum temperature increase in the tissue, SARpeak is the
maximum deposited power in the body when there is no implant in the body, α
is the diffusivity of the tissue, ct is the heat capacity of the tissue, Rb is radius of
the body, R is the radial distance from the center of the body, l is the length of
the implant lead, A is the area of the curvature of the lead, θ is the angle that
curvature of the implant makes with the radial axis, and f(Dv) is the perfusion
correction factor, which is function of the diameter of the electrode and
perfusion. The safety index formula was tested by simulations. Simulation
results showed that the theoretical safety index formula approximates and
identifies the RF heating problem of active implants with short leads accurately.
The safety index formula derived in this thesis is valid for only short
wires. However, the formulation for long wires is currently under investigation.
Despite the fact that the results obtained for short leads can not be generalized
for the safety of patients with active implants, it is believed that this study is the
first step towards safety of these patients. Using safety index as a measure of
safety is very beneficial to ensure the safety of patients with active implants.
Because, it uses the MR scanner-estimated deposited power that does not take
the existence of the implant in the patient body into account. This formulation is
the first study illustrating the advantage of the safety index metric for RF
heating studies of active implants.
Keywords: MRI, RF heating, Active Implants, RF Safety, Safety Index,
Quadrature Birdcage Coil
iv
ÖZET
VÜCUDA TAKILABİLEN TIBBİ ELEKTRONİK
ÜRETEÇLERİN MR GÖRÜNTÜLENMESİNİN
GÜVENLİK İNDEKSİ KULLANILARAK RADYO
FREKANS MODELLENDİRİLMESİ
Halise Irak
Elektrik ve Elektronik Mühendisliği Bölümü Yüksek Lisans
Tez Yöneticisi: Prof. Dr. Ergin Atalar
Ağustos 2005
MR görüntüleme güvenli bir görüntüleme tekniği olarak bilinmektedir. Fakat
kalp pili ve derin beyin uyarıcıları gibi tıbbi elektronik üreteçler taşıyan hastalar
için tehlikeli olabilir. Bunun temel nedeni elektronik üreteçlerin kablolarında
bulunan elektrotların radyo frekans (RF) dalgalar nedeniyle ısınmasıdır.
Geçmişte bu problem insan modelleri, hayvanlar ve insanlar üzerinde yapılan
deneylerle analiz edilmiştir. Ancak bu üreteçlerin elektrotlarında meydana gelen
sıcaklık artışı teorik olarak analiz edilmemiştir. Bu tezde, üreteçlerin güvenlik
indeks’ini yaklaşık olarak hesaplayan basit bir formül türetilmiştir. Güvenlik
indeks’i vücutta üreteç varken meydana gelen sıcaklık artışının, üreteç olmadığı
zaman dokuda depolanan birim enerjiye oranı olarak tanımlanmaktadır.
Bu amaçla, analitik çeyrek evre kuş kafesi sargı modeli geliştirilmiş ve
vücuda
boylamsal
düşen
elektrik
alan
dağılımı
aşağıdaki
gibi
formülleştirilmiştir:
E z ( R) = − ω µ H − R
bu formülde ω açısal frekansı, µ dokunun manyetik geçirgenliğini, H- RF
manyetik alanın sol el kuralına göre dönen bileşenini ve R vücudun
merkezinden olan radyal uzaklığı simgelemektedir. Bu formül simulasyonlar ve
v
insan modeli deneyleriyle sorgulanmıştır. Bu modelin analitik, simulasyon ve
deney sonuçları büyük oranda uyuşmaktadır.
Bu formülasyona dayanarak kısa kablolu elektronik üreteçlerin Güvenlik
İndeks (Gİ) formülü aşağıdaki gibi türetilmiştir:
SI =
∆Tmax
1
=
Rl + Ae jθ
2
SARpeak 2α ct Rb
2
f ( Dv)
Bu formülde ∆Tmax dokudaki maksimum sıcaklık artışını, SARpeak vücutta üreteç
yokken biriken maksimum gücü, α dokudaki yayılma gücünü, ct dokunun
sıcaklık kapasitesini, Rb vücudun yarı çapını, R vücudun merkezinden olan
radial uzaklığı, l üretecin kablo uzunluğunu, A kablonun eğrilik alanını, θ
üretecin kablo eğrisinin radyal eksenle yaptığı açıyı ve f(Dv) elektrot çapına ve
perfüzyona bağlı olan perfüzyon düzeltme faktörünü simgelemektedir. Güvenlik
indeks formülü simulasyonlarla test edilmiştir. Simulasyon sonuçları teorik
güvenlik indeks formülünün yaklaşık olarak kısa kablolu üreteçlerin RF ısınma
problemini açıkladığını göstermiştir.
Bu tezde türetilen güvenlik indeks formülü kısa kablolar için geçerlidir.
Uzun kablolar için olan formül üzerinde çalışmalar devam etmektedir. Kısa
kablolarla elde edilen sonuçlar üreteçleri olan hastaların güvenliği için
genellenemese de, inanıyoruz ki bu çalışma bu hastaların güvenliği için
yapılacak olan çalışmalara öncü nitelik taşımaktadır. Güvenlik indeksini
güvenlik ölçütü olarak kullanmak, MR tarayıcının tahmin ettiği vücutta üreteç
yokken depolanan gücü kullanarak hesaplandığı için elektronik üreteçleri olan
hastaların güvenliğini sağlamak açısından oldukça faydalıdır. Bu formülasyon
güvenlik indeksini güvenlik ölçütü olarak almanın ne kadar faydalı olduğunu
göstermek açısından elektronik üreteçlerin RF ısınması üzerine yapılan ilk
çalışmadır.
Anahtar Kelimeler: MR Görüntüleme, RF Isınma, Elektronik Üreteçler, RF
Güvenlik, Güvenlik İndeksi, Çeyrek Evre Kuş Kafesi Sargısı.
vi
To my father Hamdullah and Prof. Ergin Atalar
for giving me a chance.
vii
Table of Contents
1. Introduction ............................................................................................ 1
1.1 MOTIVATION AND LITERATURE SURVEY ...............................................................................1
1.2 THE OBJECTIVE AND SCOPE OF THE THESIS ...........................................................................3
2. Theory...................................................................................................... 5
2.1
2.2
2.2.1
2.2.2
2.3
2.3.1
2.3.2
2.3.3
2.4
2.4.1
INTRODUCTION ...............................................................................................................5
RF HEATING MODEL .......................................................................................................6
RF HEATING MODEL WITHOUT IMPLANTS ......................................................................6
RF HEATING MODEL WITH IMPLANTS .............................................................................8
SAFETY INDEX OF ACTIVE IMPLANTS ............................................................................9
CALCULATION OF SAR AT THE IMPLANT TIP .............................................................11
CALCULATION OF INDUCED CURRENT .........................................................................11
CALCULATION OF TEMPERATURE INCREASE IN THE TISSUE ........................................16
MODELING FIELDS INSIDE THE BODY COIL ..................................................................21
SAFETY INDEX OF ACTIVE IMPLANTS USING BODY COIL FIELD MODEL ......................27
3. Materials and Methods ........................................................................ 30
3.1
3.2
3.3
3.3.1
3.3.2
3.3.3
3.4
INTRODUCTION .............................................................................................................30
MATLAB SIMULATIONS .............................................................................................30
ELECTROMAGNETIC (EM) SIMULATIONS ....................................................................31
SIMULATION OF QUADRATURE BIRDCAGE COIL MODEL ............................................31
SIMULATION OF INDUCED CURRENT AT THE TIP .........................................................31
SIMULATION OF GEL PHANTOM ...................................................................................33
PHANTOM EXPERIMENTS .............................................................................................34
4. Results.................................................................................................... 40
4.1
4.2
4.2.1
4.2.2
4.2.3
4.3
MATLAB RESULTS........................................................................................................40
SIMULATION RESULTS .................................................................................................40
RESULTS OF QUADRATURE BIRDCAGE COIL MODEL ..................................................40
RESULTS OF INDUCED CURRENT AT THE TIP ...............................................................44
RESULTS OF GEL PHANTOM SIMULATION ...................................................................57
EXPERIMENT RESULTS .................................................................................................60
5. Discussion .............................................................................................. 64
6. Conclusion and Future Work.............................................................. 69
7. Bibliography.......................................................................................... 71
8. Appendix ............................................................................................... 74
viii
List of Figures
Figure 1. The photography of a typical active implantable medical device. The
device shown in the figure is a pacemaker (Regency SC+ 2402L,
Pacesetter, Switzerland)…………………………………………….5
Figure 2. Flow-chart model of RF heating in MRI when there is no implant in
the body. This figure is copied from reference [21]… …………….6
Figure 3. Flow-chart model of RF heating in MRI in case there exists an implant
in the body. This figure is copied from reference [6]…………….…9
Figure 4. Model of an active implantable medical device. In this figure, D is the
diameter of the tip; A is the area of the curvature; and l is the zcomponent of the distance between the metallic case and the electrode
tip………………………………………………….……………….10
Figure 5. Equivalent circuit of the active implant model. VE is the voltage source
due to the coupling of the transmitted electric field with the straight
part of the lead. VH represents the voltage source because of the
coupling of the transmitted magnetic field with the curvature of the
lead. Ztip is the impedance of the tip…………………………….....11
Figure 6. Axial view of the implant on the cylindrical human torso. I is the
induced current flowing from the metallic case through the bare tip,
that determines the direction of the normal vector nA of curvature. Φ is
the angle of the line connecting the center of the implant curvature to
the origin of the cylindrical body. β is the angle between the normal of
the loop and the x-axis. ψ is the angle that curvature of the implant
makes with the x-axis………….......................................................13
Figure 7. RF heating model of an insulated lead with bare tip, which is a very
thin wire with spherical PEC tip. When current I is injected, current
density J is distributed spherically symmetric to the tissue………..14
Figure 8. Perfusion correction factor. Although, it is a complicated function
analytically, it has a simple appearance when plotted……………..20
Figure 9. Representation of quadrature birdcage coil with plane waves…….24
ix
Figure 10. Axial view of the implant on the cylindrical human torso (see Figure
6 for a detailed description). θ is the angle that curvature of the
implant makes with the radial axis connecting the center of the
curvature with the center of the cylindrical object…………….…...27
Figure 11.Left Panel: The simplified configuration of the implant without
curvature of the lead in EM simulations. The implant was excited by
quadrature birdcage coil model. Right Panel: The spherical tip of the
implant is zoomed in………………………………………………32
Figure 12. Left Panel: The simplified configuration of the implant with only loop
in EM simulations. The implant was excited by quadrature birdcage
coil model. Right Panel: The spherical tip of the implant is zoomed
in…………………………………………………………………..33
Figure 13. Simulation of phantom inside the quadrature birdcage coil model in
the EM
Simulations……………………………………………………33
Figure 14. The envelope of the RF magnetic field……………………………37
Figure 15. The experimental setup for the phantom experiment………………38
Figure 16. Axial view of the phantom container with inserted probes into the
gel. The probes 1 and 2 are located 3 cm away from the center of the
phantom. The probes 2, 4, 7, and 9 are placed 6 cm away while
probes 5 and 6 are located 9 cm away from the center of the
phantom…………………………………………………………….39
Figure 17. The RF magnetic field at 63.8 Hz observed from the quadrature
birdcage coil model………………………………………………...41
Figure 18. The transmitted RF electric field at 63.8 Hz observed from the
quadrature birdcage coil model…………………………………......42
Figure 19. The RF magnetic field at 63.8 MHz observed from the quadrature
birdcage coil model……………………………………………...…43
Figure 20. The transmitted RF electric field at 63.8 MHz observed from the
quadrature birdcage coil model…………………………………….44
Figure 21. Induced current at 63.8 MHz as a function of wire length, l, between
the case and the bare tip when the implant is located on R = 6 cm…
x
………………………………………………………………………45
Figure 22. Induced current at 63.8 MHz as a function of wire length, l, between
the case and the bare tip when the implant is located on R = 6 cm…
…………………………………………………………………..46
Figure 23. Induced current at 63.8 MHz as a function of R when the wire length,
l, between the case and the bare tip is equal to 10cm……....…..47
Figure 24. Induced current at 63.8 MHz as a function of the diameter of the tip
when the wire length, l, is equal to 10 cm and the implant is located
at R = 6 cm……………………………………………………...48
Figure 25. Induced current at 63.8 MHz as a function of the conductivity of the
medium when the wire length, l, is equal to 10 cm and the implant is
located at R = 6 cm……………………………………………...49
Figure 26. Induced current at 63.8 MHz as a function of R when the area, A, of
the curvature is equal to 20 cm2 when there is no wire between the
case and the tip. ............................................................................... 50
Figure 27. Induced current at 63.8 MHz as a function of area, A, of the curvature
when there is no wire between the case and the tip. The center of the
curvature is located at R = 6 cm. ....................................................... 51
Figure 28. The modified version of implant configuration in Figure 11,
configuration with only curvature of the wire when l = 0…………..52
Figure 29. Induced current at 63.8 MHz as a function of R when the case was
placed with a wire connected to a sphere, and the area, A, of the
curvature is equal to 20 cm2. ............................................................ 53
Figure 30. Induced current at 63.8 Hz as a function of R when the case was
placed with a wire connected to a sphere, and the area, A, of the
curvature is equal to 20 cm2. ............................................................ 53
Figure 31. Induced current at different frequencies as a function of R when the
case was placed with a wire connected to a sphere, and the area, A,
of the curvature is equal to 20 cm2. The results were normalized with
respect to the 63.8 MHz data............................................................ 54
Figure 32. Induced current at 63.8 MHz as a function of area, A, of the curvature
the case was placed with a wire connected to a sphere. The center of
the curvature is located at R = 6 cm. ................................................. 55
Figure 33. The most primitive version of implant configuration in Figure 11,
xi
configuration with a wire of loop that shown segment was loaded
with the impedance of the tip. .......................................................... 56
Figure 34. Induced current on a loop of wire whose one of the segments loaded
with the impedance e of the tip as a function of R. the area of the loop
is equal to 20 cm2. ............................................................................. 57
Figure 35. Transmitted electric field distribution inside a cylindrical gel
phantom at the center, and the height of 5 cm, 10 cm, 15 cm and 20
cm. .................................................................................................... 58
Figure 36. One of the typical experimental data. This is the data of the
experiment 2 with probes 3, 4, 5 and 6. ........................................... 61
Figure 37. Two different configurations may result in significantly different
heating at the tip. Right panel shows the maximum heating
configuration. ................................................................................... 67
xii
List of Tables
Table 1. The conductivity and electrical permittivity measurements of prepared
gel ....................................................................................................... 35
Table 2. Dielectric properties of the human heart, brain and blood [38]............35
Table 3. Thermal properties of the human heart, brain and blood……………..36
Table 4. Estimated errors of induced current at the tip at different frequencies
when the case was placed with a wire connected to a sphere, and the
area, A, of the curvature is equal to 20 cm2......................................... 54
Table 5. The transmitted field results of gel phantom inside the quadrature
birdcage coil model in the EM simulation. However, SAR is calculated
by hand using Eq.(1) ........................................................................... 59
Table 6. The calculated transmitted field and resulted SAR values for the gel
phantom inside the quadrature birdcage coil model............................ 59
Table 7. The estimated error between the Electromagnetic Simulation data and
the theoretical data............................................................................... 60
Table 8. The calculated theoretical SAR values................................................. 60
Table 9. Results of experiment 1 with probes 1, 2, 5, and 6 .............................. 61
Table 10. Results of experiment 2 with probes 3, 4, 5, and 6 ............................ 62
Table 11. Results of experiment 3 with probes 1, 2, 3, and 4 ............................ 62
Table 12. Results of experiment 4 with probes 2, 3, 4, and 6 ........................... 63
Table 13. Results of experiment 5 with probes 4, 7, 8, and 9 ............................ 63
Table 14. Results of experiment 6 with probes 3, 4, 7, and 8 ........................... 63
xiii
Chapter 1
Introduction
1.1 Motivation and Literature Survey
There is an increasing demand for MRI exams of patients with active
implantable medical devices (AIMD) such as pacemakers and deep brain
stimulators. Unfortunately, these patients are barred from MRI exams primarily
due to the possibility of hazardous RF heating at the lead tips of the implants.
Despite the fact that most of the studies on the RF heating of implants are
limited by testing of leads with phantom [1-3], animal [4] and human [5]
experiments, there are a small number of quantitative studies [6-8]. These
assessments focus on the heating of a straight wire as a good approximation for
the RF heating of an implant with insulation [6], without insulation[6, 7], and
insulation with exposed tips [7, 8]. Furthermore, the effects of loops constructed
by leads were computed [9] and observed experimentally with phantom
experiments [3, 10-12]. In the literature, there is no study with the aim of finding
an analytical formula for the problem of AIMD heating. Such formula will
enable us 1) to understand the parameters affecting the tip heating; 2) to
determine the maximum power that can be safely applied during an MRI
examination; and 3) to develop novel and effective methods to reduce coupling
between AIMD and the MRI scanner.
It was reported [13] that just after an MRI examination of the head at 1
Tesla (T), a 73-year-old patient with bilateral implanted deep brain stimulator
electrodes for Parkinson disease showed dystonic and partially ballistic
movements of the left leg. Despite the fact that MR imaging of patients with
deep brain stimulators was performed many times at 1.5 T with no side effects
1
before this incident, this incident shows that the generalization of the same
conditions even at lower field strengths, i.e. 1 T, can be dangerous.
Consequently, it is suggested [14-17] that each MRI system and update of the
same system needs a specific safety regulation assessed with a preclinical study
for the RF heating of the specific implant leads during MRI examinations.
According to these studies [14-18], since MRI scanner calculated SAR does not
take the existence of the active implants in the body into account, it is not
reliable to use it for ensuring the safety of patients. Therefore, safety
recommendations developed for a certain system, i.e. type of implant, body coil,
MRI system and field strength, especially when the estimated SAR of the
system is concerned may not be implemented across different MRI systems.
As it is seen, there is a doubt in the literature for the reliability of SAR
(when there is an implant, people do not speak much on SAR but they focus on
temperature) for the RF safety of patients with active implant. All of these
controversies imply that all parameters of the RF heating problem should be
clearly determined and put in a comprehensible form such that it becomes easier
to develop universal RF safety limits. To achieve such practical format, the RF
heating of active implants should be analytically analyzed. Considering the
variety of the MRI scanners and active implants, and how regularly they are
updated, it is more realistic to search for a consistent solution in which the type
of the MRI system is not a parameter.
In this study, the MRI-related RF heating problem of active implantable
medical devices (AIMDs) with a single short lead considering the curvature of
the lead was theoretically analyzed at 1.5 T. The safety index [6] of active
implants; the temperature rise at the implant lead tip per unit deposited power in
the tissue without the implant in place, was also formulated. For that purpose, an
active implant was placed in an infinitely long cylindrical human body model. It
is assumed to lie coaxial with the transmit body coil. Then, the incident electric
field on the implant was analytically found under the quasistatic assumption.
2
Next, the induced current, the SAR amplification and the resultant temperature
increase in the tissue was formulated in terms of the electrical and thermal
characteristics of the tissue and the characteristics of the implant configuration
such as radius of the tip, length of the lead, area of the lead curvature and the
position of the implant. Finally, the resultant temperature at the tip was
normalized with the peak SAR in the body to find the safety index of the active
implant. As a result of this analysis, it was aimed to explain how important and
practical to use safety index metric in order to ensure the RF heating safety of
active implants. Thus, this study shows that RF heating can be analytically
identified and the safety index formula helps the standardization of the MRIrelated RF heating problem. Besides, since it is required to calculate the safety
index, scanner estimated SAR was shown to be vital for the RF safety of
patients with active implants.
1.2 The Objective and Scope of the Thesis
In this thesis, an analytical analysis of the RF heating problem of AIMDs was
done with the purpose of deriving a general formulation for the safety index of
AIMDs. After calculating the induced electrical field on the cylindrical human
model, the amplified absorbed power at the tip of the AIMD was found. Next,
the temperature increase in the tissue was studied and finally the safety index of
the AIMD was calculated by normalizing the temperature rise with respect to the
absorbed power in the tissue when the AIMD is not in place.
This thesis has been divided into six chapters. Chapter 1 is devoted to
the introduction and motivation. Chapter 2 explains the RF heating model of
AIMDs during MRI scans with its implementation on an AIMD inside a
cylindrical human body model under the quasistatic assumption. Chapter 3
contains the materials and methods used for the testing of the analysis. Chapter 4
is devoted the obtained results whereas Chapter 5 goes into the discussions of
the results. Finally, Chapter 6 includes the conclusions of the thesis with the
future work.
3
In the appendix, additional information about the Fourier transform
convention and the special integral identities used during the calculations are
presented.
4
Chapter 2
Theory
2.1. Introduction
A typical active implantable medical device (AIMD) with a single lead is shown
in Figure 1. In order to analyze the RF heating of these types of implants, a
general configuration was developed. The length of the lead was shortened
despite the fact that it is very long. Also, the lead was considered as curved
rather than looped around itself.
Figure 1. The photography of a typical active implantable medical device. The
device shown in the figure is a pacemaker (Regency SC+ 2402L, Pacesetter,
Switzerland).
In that chapter, first an early proposed RF heating model was explained.
Also, the safety index of an AIMD was calculated based upon that model with
underlined simplifying assumptions. The same derivations for the safety index
were overviewed and simplified with a novel quadrature birdcage body coil
model.
5
2.2. RF Heating Model
When a body undergoes an MRI examination, the body heats up due to the RF
fields transmitted by the MRI scanner. The underlying mechanism behind the
RF heating was modeled by Bottomley et al [19] and formulated what happens
when there is an implant by Yeung et al [6]. This model was developed at 1.5 T
for two cases; when there is an implant in the body and there is not an implant in
the body. The detailed explanations are given in the following subsections.
2.2.1. RF Heating Model without Implants
RF heating of a body without an implant is the most common case. The body is
exposed to the RF power of a transmitting body coil when there is no metallic
implant in the body [20]. As shown in Figure 2, P is the time-averaged input
power and determined by the applied RF pulse of the imaging sequence. This
input power causes power deposition, characterized by specific absorption rate
G
(SAR) and a function of position r , in the body with respect to the
electromagnetic properties of the tissue. Afterwards, this SAR distribution is
G
converted into temperature distribution, which is a function of position r as
well, depending on the thermal properties of the body [21].
P
Transmit
Coil
G
SAR(r)
Bioheat
Transfer
G
T(r)
Figure 2. Flow-chart model of RF heating in MRI when there is no implant in
the body. This figure is copied from reference [21].
The SAR is calculated from the electric field distribution in the body
according to the following equation:
SAR =
σ
| E |2
ρt
(1)
6
where σ is the electrical conductivity of the tissue , ρt is the mass density of the
tissue, and E is the rms amplitude of the RF electric field transmitted by the
body coil. This can be calculated by using Maxwell’s equations.
The temperature distribution in the body can be calculated by the
bioheat equation first proposed by Pennes [22].
α ρb cb m
dT (r , t )
1
= α ∇ 2T (r , t ) −
(T (r , t ) − Tb ) + SAR(r , t ) + Q
dt
ct
ct
(2)
in which ct and α are the heat capacity and thermal diffusivity of the tissue
respectively, cb, ρb and Tb in that order are the heat capacity, mass density and
temperature of the perfusing blood, m is the volumetric flow rate of blood per
unit mass, Q is the heat generated by normal chemical processes in the body, ∇
is the Laplacian operator, and r is the position vector. When it is assumed that
metabolic heat generation keeps the core body temperature steady with the
perfusing blood temperature [21], the Eq.(2) can be written as follows:
d ∆T (r , t )
1
= α ∇ 2 ∆T (r , t ) − α v 2 ∆T (r , t ) + SAR(r , t )
dt
ct
(3)
where ∆T = T - Tb and v, defined as v = ρ b cb m / α ct , is the lumped perfusion
constant. Notice that with that assumption the effect of the thermoregulation
constant, Q, is assumed to be zero.
Even if the analytical solution of Eq. (3) is not possible, in case of local
heating, it is possible to achieve an approximate solution with the following
simplifying assumptions. First, the thermal parameters are assumed to be
constant around the point of interest over a small temperature range. Next, the
local region is assumed to be small with respect to whole body and not near the
surface of any boundary. With these assumptions, Green’s function of the
bioheat equation can be used to find the spatial temperature distribution [21] as
the following equation:
∆T (r ) = SAR(r )* G (r )
(4)
7
where ‘*’ denotes convolution and G (r ) is the Green’s function of the bioheat
equation as a function of position vector.
Green’s function of the bioheat equation in the cylindrical (line source)
and spherical (point source) coordinates for steady state are given respectively in
Eq. (5) and Eq.(6).
G ( R) =
G (r ) =
1
2πα ct
K 0 ( vR )
(5)
1 e − vr
α ct 4π r
(6)
where R is the distance from line source, r is the distance from point source, v is
a lumped perfusion parameter, α is the diffusivity of the tissue, ct is the heat
capacity of the tissue, and K0 is the modified Bessel function of the second kind
and order zero.
The time-dependent Green’s function of the bioheat equation in
cylindrical in and spherical in coordinates are given in [21].
2.2.2. RF Heating Model with Implants
RF heating of a body with an implant is a very complex case because of the
coupling of the transmitting coil with the metallic implant [20]. When there is an
implant in the body during an MRI procedure, the absorbed power (SAR) in the
body is amplified around the implant. This amplification is quantified with SAR
gain as shown in Figure3A. With current technology it is not possible to know
the resultant amplified SAR, denoted SAR’ in Figure3A in the body. Yet, the
deposited power when there is no implant in the body, that is SAR, can be
estimated because the current MRI scanners are designed in a way that the
applied power level is always lower than the patient safety limits. Therefore, it is
reasonable to combine the raw SAR gain with the bioheat transfer into one unit,
safety index [6], so that we can have a system with an estimated input, SAR, in
8
order to be able find the resultant temperature distribution in the body as shown
in Figure3B. This helps to ensure safety of patients with implants by setting
limits on the applied SAR of the imaging pulse sequence.
Thus, the safety index expresses the temperature increase as a
consequence of the existence of an implant in the body for each unit of peak
applied SAR when there is no implant in the body.
A
B
P
Transmit
Coil
P
Transmit
Coil
G
SAR(r)
SAR
Gain
G
SAR(r)
G
SAR′(r)
Safety
Index
Bioheat
Transfer
G
T(r)
G
T(r)
Figure 3. Flow-chart model of RF heating in MRI in case there exists an
implant in the body. This figure is copied from reference [6].
2.3. Safety Index of Active Implants
With the proposed RF heating model for the existence of an implant in the body
exposed to MRI-related RF fields, in that section, the safety index of active
implants is obtained analytically by implementing each system in the RF heating
model.
The analysis for the coupling of the transmit coil with a metallic implant
in the body during an MRI examination is rather complicated despite the fact
that it is straightforward for an electromagnetic solver software to analyze.
Therefore, the theoretical analysis of the implant lead tip heating problem is not
possible without some simplifying assumptions.
9
In order to make the first order approximation for the RF heating of an
AIMD lead tip, the human body was assumed to be an infinitely long cylindrical
object with uniform electromagnetic properties as in earlier studies [19, 23-25].
It was assumed to be lying coaxial with the transmit body coil. Besides, similar
to these studies, the diameter of the body was assumed to be small compared to
the wavelength.
On the other hand, considering the variety of AIMD configurations, the
implant in Figure 1 was simplified for the analysis as shown in Figure 4. The
common structure of an AIMD includes at least one insulated lead with a bare
lead tip and a generator with a metallic case. Besides, the size of the implant
including its leads was assumed to be significantly smaller than the wavelength;
hence quasistatic RF fields can be used around the implant during the analysis.
metallic
case
Figure 4. Model of an active implantable
medical device. In this figure, D is the
diameter of the tip; A is the area of the
curvature; and l is the z-component of the
distance between the metallic case and
the electrode tip.
A
l
insulated
lead
bare
tip
D
10
2.3.1. Calculation of SAR at the Implant Tip
The transmitted RF magnetic and electric fields of the body coil are coupled
with the active implant in a way that a potential difference between the metallic
case and the bare tip is induced. This gives rise to a current induction on the
implant lead. This current is scattered from the bare tip to the tissue. Since tissue
is a lossy medium, this scattered current amplifies the absorbed power already
induced by the transmitted RF magnetic field in the body.
2.3.1.1. Calculation of Induced Current
There are two sources of voltage induction between the metallic case and the
bare tip; the transmitted RF magnetic field and electric field. The resultant
potential difference gives rise to a current induction at the lead tip of the implant
depending on the impedance of the tip in a way that will be explained next. The
equivalent circuit of an implant shown in Figure 5 illustrates the parameters
affecting the current induction at the tip.
VE
+
_
I
Zt i p
VH
+
_
Figure 5. Equivalent circuit of the
active implant model. VE is the
voltage source due to the coupling
of the transmitted electric field
with the straight part of the lead.
VH represents the voltage source
because of the coupling of the
transmitted magnetic field with
the curvature of the lead. Ztip is
the impedance of the tip.
The first source comes from the coupling of the straight part of the lead
with the incident electric field. It is calculated as follows:
Ve = Ez ( R) ⋅ l
(7)
11
where Ez(R) is the transmitted electric field in longitudinal direction and varies
only in the radial direction on the body, and l is the distance between the
metallic case and the tip in z-direction. In most of the studies calculating the
deposited power around a straight wire [6, 7]; the maximum coupling between
the electric field and the straight wire was achieved when the electric field is
incident parallel to the wire. In our case, the transmitted electric field was also
parallel to the straight part of the lead for maximum coupling. Then, Eq.(7) is
reduced to the following:
Ve = Ez ( R)l
(8)
Reilly and Diamant [26] used the same idea for the excitation of a nerve fiber by
an external electric field to analyze the peripheral nerve stimulation.
On the other hand, the second source comes from the coupling of the
curvature constructed by the lead with the RF magnetic field. This source of
voltage induction can be calculated by Maxwell’s Faraday’s law of induction as
shown below:
Vh = − jω B1.nA A
(9)
in which A is the area of the loop and nA is the normal vector of the loop and B1
is the RF magnetic field. In fact, in many studies [3, 9-11, 27], it is mentioned
that the effect of the existence of loops or curvature of leads can be calculated as
shown in Eq. (9). Here, B1 can be written in vector form depending on the lefthand rotation (this is the only one exciting the spins) convention as follows:
B1 = B1(aˆ x + jaˆ y )
(10)
For a general analysis of the implant structure, it was assumed that the normal of
the loop makes an angle of β with x-axis as shown in Figure 6; therefore, it can
be expressed in vector form as follows:
nA A = aˆ x Acosβ + aˆ y Asinβ
(11)
12
Figure 6. Axial view of the implant on
nA
y
the cylindrical human torso. I is the
induced current flowing from the
β
I
metallic case through the bare tip, that
ψ
determines the direction of the normal
vector nA of curvature. Φ is the angle of
the line connecting the center of the
implant curvature to the origin of the
R
cylindrical body. β is the angle between
the normal of the loop and the x-axis. ψ
φ
x
is the angle that curvature of the
implant makes with the x-axis.
Thus, the resultant induced voltage due to the curvature of the leads can be
written as the following:
Vh = − jω B1 Ae jβ
(12)
As a result, the total induced voltage between metallic case and the tip
can be written as follows:
Vtotal = Ez ( R)l − jω B1 Ae jβ
(13)
We can write Eq. (13) in a more appropriate form as follows:
Vtotal = Ez ( R )l + ω B1 Ae jψ
(14)
in which ψ is defined as ψ = β − π / 2 shown in Figure 6.
13
Figure 7. RF heating model
of an insulated lead with bare
tip, which is a very thin wire
with spherical PEC tip.
When current I is injected,
current density J is
distributed spherically
symmetric to the tissue.
So as to find the induced current, call that current I, at the tip using the
voltage difference between the bare tip and the metallic case, the impedance of
the body to the tip should be calculated. For that purpose, the potential of the tip
with respect to the lossy medium around it can be calculated by assuming a
current I is induced at the tip. For simplicity, the metallic case and the tip were
assumed as perfect electric conductor (PEC). While the insulated wire was
assumed to be very thin, the tip was assumed to be a sphere with diameter D in
order to take advantage of the spherical symmetry of the tip as shown in Figure
7. This assumption enabled to treat the bare tip as a point source, scattering
current density around it. Thus, it became possible to use Green’s function for a
point source of the bioheat equation to calculate the heat transfer to the tissue.
Assuming the bare tip as a PEC enabled the worst-case heating, because the
small resistivity of the tip gives rise to maximum SAR amplification at the tip
[28].
In a lossy medium, there are two types of current density in the system;
conduction and displacement current densities as giving in Eq.(15). Due to the
spherical symmetry, the current density is defined as in Eq. (16). Next, the
scattered electric field can be found as in Eq. (17).
J = (σ + jωε ) E
(15)
14
J=
I
(16)
4π r 2
Er =
Ι
(17)
(σ + jωε ) 4π r 2
Then, we can find the potential of the tip by integrating the scattered
electric field with respect to the radial distance to the tip as shown below:
D /2
Vtip = − ∫ E ⋅ dr
(18)
∞
Thus, the potential of the tip becomes the following:
Ι
Vtip =
2π (σ + jωε ) D
(19)
Then, the impedance of the tip modeled as a spherical PEC in a dielectric
medium can be formulated as follows:
Z tip =
1
2π (σ + jωε ) D
(20)
Notice that in Eq. (19) the potential of the tip decreases as the diameter of the
tip increases. Therefore, we can consider the metallic case with a very large
diameter so that its potential with respect to tip becomes negligible. Then, the
voltage difference between the tip and the case can be approximated as the
potential of the tip with respect to the dielectric medium around it. Then, the
induced current at the tip can be found as the following:
Ι = 2π (σ + jωε ) D (Ez ( R)l + ω B1 Ae jψ )
(21)
Using induced current on the lead, we can formulate the scattering
electric field at the tip and consequently the amplified SAR at the tip. Thus,
from Eq.(17), the induced electric field at the tip can be formulated in terms of
the incident RF fields and the properties of the active implant as follows:
Er ( r ) =
D
E ( R)l + ω B1 Ae jψ )
2( z
2r
(22)
Next, the amplified deposited power at the tip of the implant can be
calculated as a function of radial distance r as follows:
15
2
SAR′(r ) =
σ⎛ D ⎞
jψ 2
⎜ 2 ⎟ Ez ( R )l + ω B1 Ae
ρ t ⎝ 2r ⎠
(23)
Then, maximum amplified power can be written like this:
2
2
σ 2
′ = SAR′( r = D/2) = ⎛⎜ ⎞⎟ Ez ( R )l + ω B1 Ae jψ
SARmax
ρt ⎝ D ⎠
(24)
2.3.2. Calculation of Temperature Increase in the
Tissue
With the known amplified SAR distribution, the temperature rise distribution
can be calculated in terms of induced current I at the tip using Green’s function
averaging technique [21] to take into account the bioheat transfer effects.
Notice that, in this study only the temperature increase due to the
existence of the implant, in other words the amplified SAR (SAR’) at the tip will
be calculated. Even if there is no implant in the body, body temperature
increases due to the deposited power (baseline SAR) during an MRI procedure.
However, this increase is kept limited by the MRI system, i.e. pulse sequences
are adjusted in a way that the applied power is always lower than the safety
limits. Therefore, the temperature rise due to the baseline SAR is not a safety
concern and much lower compared to the one caused by the deposited power
due to the existence of an implant (SAR’). SAR’ causes such a high temperature
increase at the tissue that the tissue might burn. Thus, only the effect of scattered
electric field on the temperature rise in the tissue was calculated by neglecting
the one cause by the transmitted electric field.
From Eq. (23) , the amplified SAR at the tip can be formulated as
follows:
16
⎧ σ ⎛ D ⎞2
jψ
⎪ ⎜ ⎟ Ez ( R )l + ω B1 Ae
⎪ ρt ⎝ 2 ⎠
SAR′(r ) = ⎨
D
⎪
for r <
⎪⎩0,
2
2
1
,
r4
for r >
D
2
(25)
In steady-state, the Green’s function of the tissue bioheat equation in
spherical coordinates [21] for a point source is given as:
G (r ) =
1 e − vr
α ct 4π r
(26)
where ct and α are the heat capacity and thermal diffusivity of the tissue
respectively and v is a lumped perfusion constant and r is the distance from point
source. Lumped perfusion constant is defined as v = ρ b cb m / α ct , in which
ρ b and cb are in that order the mass density and heat capacity of blood, and m is
the volumetric flow rate of blood per unit mass of tissue [21].
Then, the amplified SAR can be convolved with the Green’s function of
the bioheat equation as it is given in Eq. (4). However, a different methodology
suggested by Gao et al. [29] was followed for our calculations. That is, while
calculating the resulting temperature distribution, the Fourier transform and its
properties like convolution property given in Eq.(83) were used for simplicity
rather than using convolution in computations directly. Thus, the Fourier
transform of the amplified SAR was multiplied with the Fourier transform of the
Green’s function of the bioheat equation, and then their spherical inverse Fourier
transform was taken to find the temperature increase.
The Fourier transform of spherically symmetric SAR′(r ) distribution
was calculated using (81) in this way:
2
2 4π
σ ⎛D⎞
SAR′(q ) = ⎜ ⎟ Ez ( R )l + ω B1 Ae jψ
ρt ⎝ 2 ⎠
q
∞
sin(qr )
dr
r3
D
∫
2
17
(27)
The spherical Fourier transform (81) of Green’s function can be found
using the identity in Eq. (84) in this way:
G (q ) =
1
1
2
α ct v + q 2
(28)
Using (82) and (83),
∆T ( r ) =
∞
1
(2π )
3
∫ SAR′(q)G(q)
0
sin(qr )
4π q 2 dq
qr
(29)
Yet, r in SAR′(q ) is not the same as r in ∆T ( r ) , so the notation of SAR′(q ) was
changed as r ′ .
σ
∆T ( r ) =
ρt
⎧
⎫
2
∞ ∞
⎪ 1
⎛D⎞
jψ 2 1 2 1 ⎪ sin( qr ′)
dr ′⎬ 2
sin(qr )dq
⎨∫
⎜ ⎟ Ez ( R )l + ω B1 Ae
3
2
∫
α ct π r 0 ⎪ D r ′
⎝2⎠
⎪v +q
⎩2
⎭
(30)
Here, a trick was made by changing the order of integrals.
2
∞
∞
2 1 2 1
σ ⎛D⎞
1 ⎧ sin(qr ′) sin(qr ) ⎫
∆T (r ) = ⎜ ⎟ Ez ( R)l + ω B1 Ae jψ
dq ⎬dr ′
⎨
ρt ⎝ 2 ⎠
α ct π r ∫D r ′3 ⎩ ∫0
v2 + q2
⎭
2
(31)
Using the trigonometric identity 2sin( A)sin( B) = cos( A − B) − cos( A + B)
2
∞
∞
2 1 2 1
σ ⎛D⎞
1 ⎧ cos(q | r ′ − r |) − cos(q (r ′ + r )) ⎫
dq ⎬dr ′
∆T (r ) = ⎜ ⎟ Ez ( R )l + ω B1 Ae jψ
⎨
ρt ⎝ 2 ⎠
α ct π r ∫D r ′3 ⎩ ∫0
2(v 2 + q 2 )
⎭
2
(32)
Using the integral identity in Eq. (85), the temperature increase can be written as
follows:
⎧
2
⎫
∞
∞ − ( r ′+ r ) v
2 1
σ ⎛D⎞
1 ⎪ e −|r ′− r|v
e
⎪
′
∆T (r ) = ⎜ ⎟ Ez ( R )l + ω B1 Ae jψ
−
dr
dr ′⎬
⎨∫
3
3
∫
′
′
ρt ⎝ 2 ⎠
α ct 2rv ⎪ D r
r
D
⎪
⎩2
2
⎭
(33)
After using the definition of the absolute value,
18
σ
∆T ( r ) =
ρt
⎧r
⎫
2
∞ − ( r ′− r ) v
∞ − ( r ′+ r ) v
1 ⎪ e ( r ′− r ) v
e
e
⎪
⎛D⎞
jψ 2 1
dr ′ + ∫
dr ′ − ∫
dr ′⎬
⎨∫
⎜ ⎟ Ez ( R )l + ω B1 Ae
3
3
3
α ct 2rv ⎪ D r ′
r′
r′
⎝2⎠
D
r
⎪
⎩2
⎭
2
(34)
When r =
D
, maximum temperature increase on the tip was obtained. Then, the
2
first integral in Eq. (34) disappears. After some arrangements in the arguments
of exponentials by taking the common term Dv/2 out of the parenthesis, the
following equation can be observed:
⎧
2
∆Tmax
−(
2r′
−1)
∞
σ ⎛D⎞
1 ⎪ e D
jψ 2 1
= ⎜ ⎟ Ez ( R)l + ω B1 Ae
⎨
ρt ⎝ 2 ⎠
α ct Dv ⎪ ∫D r ′3
Dv
2
∞
dr ′ − ∫
e
−(
r ′3
D
2
⎩2
2r′
Dv
+1)
2
D
⎫
⎪
dr ′⎬
⎪
⎭
(35)
By making a change of variable
2r ′
D
= r ′′ => dr ′ = dr ′′
2
D
⎧
∆Tmax
− ( r ′′ −1)
∞
2
σ
1 2 ⎪ e
Ez ( R )l + ω B1 Ae jψ
=
⎨
ρt
2α ct Dv ⎪ ∫1 r ′′3
⎩
Dv
2
∞
dr ′′ − ∫
1
e
− ( r ′′ +1)
r ′′3
Dv
2
⎫
⎪
dr ′′⎬
⎪
⎭
(36)
Here, we call the numerical integral part of that expression Perfusion
Correction Factor that is plotted in Figure 8.
Dv
⎧ ∞ − ( r ′′−1) Dv2
⎫
∞ − ( r ′′+1) 2
2 ⎪ e
e
⎪
f ( Dv) =
dr ′′ − ∫
dr ′′⎬
⎨∫
3
3
Dv ⎪ 1 r ′′
r ′′
1
⎪
⎩
⎭
19
(37)
Figure 8. Perfusion correction factor. Although, it is a complicated function
analytically, it has a simple appearance when plotted.
Then, maximum temperature rise can be written as the next:
∆Tmax =
σ
E ( R )l + ω B1 Ae jψ
ρt z
2
f ( Dv)
2α ct
(38)
From Eq. (21),
∆Tmax =
| Ι |2
σ
1 1
f(Dv)
2
2
2 2
8π ( σ + ω ε ) αct ρt D 2
(39)
That formulation illustrates the effect of induced current at the tip to the
temperature increase in the tissue. Temperature rise can be limited by the
induced current at the tip.
Finally, the safety index can be found as follows;
Ez ( R)l + ω B1 Ae jψ
∆Tmax
=
SI =
2
SAR peak
Ez ( Rb )
2
f ( Dv)
2α ct
20
(40)
2.4. Modeling Fields inside the Body Coil
One of the main hardware components of an MRI scanner are RF coils. Their
main function is to transmit the RF magnetic field B1 homogeneously to the
body in order to excite the spins. The spins are maintained in equilibrium by the
static magnetic field B0 before the RF magnetic field excites them to the plane
perpendicular to B0 field. Then, an MR signal is obtained. The reception of this
MR signal is performed by RF coils as well.
However, the design of an RF coil with homogeneous transmission is a
very challenging task and another research area. It is desirable to produce a high
signal-to-noise ratio (SNR) in MRI systems to obtain a better spatial resolution
in the images. SNR is defines as the B1 field produced per unit coil current and
tissue losses. This is succeeded with high static field intensity. As the intensity
of B0 field increases, the frequency of the B1 field also increases. Considering
the wavelength of the B1 field, at higher frequencies the portion of the biological
body to be imaged can be comparably large or even larger than the wavelength.
Thus, this yields a stronger interaction between the biological tissues and the
electromagnetic field. This interaction not only corrupt the B1 field
homogeneity; consequently causing low quality images, but also causes more
power deposition in the body resulting safety problems for the MRI systems.
Concerning the safety issues, the high frequency and increased B1 field
homogeneity introduces much intense electric field and accordingly eddy
currents. These currents give rise to increased specific absorption rate (SAR)
which is the primary source of temperature increase in the tissue [30].
Ever since they were introduced in 1985 [31], birdcage coils have been
extensively used as a body coil in MR scanners due to their high RF field, B1,
homogeneity and high SNR over a large volume inside the coil. Homogeneity of
RF field inside a birdcage coil is proportional to the number of excitations
constructed at the corresponding legs of the coil. With birdcage coils, two types
21
of excitation are possible; linear excitation and quadratic excitation. Linear
excitation gives rise to a linearly polarized B1 field when it is fed at a single
point while quadratic excitation produces a circularly polarized B1 field when it
is fed at two points perpendicular to each other with a phase difference of 90°
[32]. Quadrature birdcage coils have more advantages than linear ones in terms
of the excitation power with 50% reduction and SNR with
2 times
amplification [23]. Therefore, quadrature birdcage coils are widely used in the
current MRI systems.
The safety problem of electromagnetic interaction with the human body
during MRI scans has been studied comprehensively for many years. Earlier
works used an approximate model of human body, an infinitely long
homogeneous cylindrical body [19, 23-25], and the resultant analytical solution
of the problem showed that as the frequency of B1 field increases, the
homogeneity of B1 field decrease while SAR in the object increases as well.
With the advance of computer processor speed and electromagnetic simulators,
it becomes possible to solve Maxwell’s equations accurately on the two
dimensional (2D) [32] and three dimensional (3D) [30, 33, 34] models of human
body. Despite the fact that these numerical solutions provide a beneficial insight
of the safety problem of MRI (especially at high frequencies between 64 MHz
and 300 MHz) and validate the results of analytical solutions, they are expressed
in complicated expressions, but make sense with their simulations.
Since the purpose in that study is to develop an analytical study to derive
a simple and easy to use safety index formula, an analytical relationship between
the transmitted electric field and RF magnetic field is needed. Because of this,
we analytically calculated the RF magnetic field, B1, over an approximate
human body model, an infinitely long cylinder with homogeneous electrical
properties, inside a quadrature birdcage coil. Since most of the MRI systems use
circularly polarized quadrature birdcage body coil due to its high field
22
homogeneity, we developed this model for quadrature birdcage body coil to
generate a homogeneous transverse RF magnetic field in the human subject.
In order to analyze the safety index of an active implant that is
absolutely independent of the MRI system used, phase distribution of the
transmitter must be carefully adjusted so that worst-case tip heating is
guaranteed [35]. Thus, the quasistatic MRI fields for the analysis were assumed,
so that the phase of the fields varies slowly on the implant that is small in
comparison with the wavelength; therefore, constructive addition of fields at the
tip was enabled for maximum heating. This phase distribution is even worse
than the worst case heating distribution mentioned in [35]. As a result, plane
waves can be used as a source of excitation. Then, assuming a homogeneous RF
magnetic field, B1, inside the body coil, the incident electric field on the implant
with a slowly varying linear magnitude and worst-case phase distribution can be
derived as a function of position.
For circularly polarized plane waves, at least two linearly polarized
plane waves with a 90º phase difference are needed. As the simplest
approximation for the birdcage coil, four plane waves were used considering the
circular geometry of the coil with equal magnitudes and appropriate phases to
achieve circular polarization. In fact, infinitely many excitations would be
ideally more accurate for the modeling of a birdcage coil. Yet, for the ease of
calculation, four plane waves were enough for the purpose.
As a background, it should be kept in mind the following constitutive
relation;
B1 = µ H 1
(41)
in which B1 is the magnetic flux density and H1 is the magnetic field intensity.
While the main magnetic field, B0, is in the z-direction, the RF magnetic
field, B1, has only transverse components and they can be written as a sum of
23
the left and right hand circularly polarized rotating field components as in the
Eq. (42).
H 1 = aˆ x H x + aˆ y H y = aˆ − H − + aˆ + H +
(42)
where â− is the left-hand sense (clockwise) rotation axis and â+ is the righthand sense (counter-clockwise) rotation axis, which are defined as follows:
aˆ − = aˆ x + jaˆ y
and
H − = ( H x − jH y ) / 2
(43)
aˆ + = aˆ x − jaˆ y
and
H + = ( H x + jH y ) / 2
(44)
in which we can estimate the magnitude of magnetic field intensity, H1, from
the imaging parameters, i.e. flip angle, pulse sequence, TR.
As a rotating frame of reference, left-hand sense rotating frame was
assumed because it is the only component that produces a torque on the
magnetization of hydrogen spins at the larmor frequency, while the other
component has no effect.
Consequently, the formulation was based on the mentioned assumptions
and references for the construction of plane waves in Figure.
z
z
z
E1z
+ H2x
x
x
H1x
y
+
E
2
z
z
E4z
z
+
y
=
x
3
y
4
y
H
y
H-
H
x
E 3z
y
x
Figure 9. Representation of quadrature birdcage coil with plane waves.
24
y
We can write the wave equations of plane waves at the center of the
body with the same amplitude considering their propagation directions as
follows:
H 1x = aˆ x
H − − jkc y
e
2
(45)
H x2 = aˆ x
H − jkc y
e
2
(46)
H y3 = j aˆ y
H − − jkc x
e
2
(47)
H y4 = j aˆ y
H − jkc x
e
2
(48)
Then, the total magnetic field intensity gives the following:
H1 = H − ( aˆ x cos ( kc y ) + jaˆ y cos ( kc x ) )
(49)
where H − is the magnetic field intensity, aˆ x and aˆ y are the unit vectors, x and y
are the distance variables on the corresponding cartesian coordinates, and j is
the complex number, j = −1 a and kc is the complex wavenumber. For circular
polarization, H − = H 10 (1 − j ) / 2 . The complex wave number is calculated [36]
as k c = ω 2 µε − jωµσ ,in which ω is the angular frequency, and µ, ε and σ
are respectively the magnetic permeability, electrical permittivity and
conductivity of the tissue. Notice that for small kc and distance from the center
of the cylindrical human body, Eq. (49), gives the next:
H1 = H − ( aˆ x + jaˆ y )
(50)
that is consistent with the definition of left hand circularly polarized magnetic
field intensity.
With that excitation scheme, left hand circularly polarized magnetic field
was ensured only at the center of the body.
25
Next, from Maxwell’s Ampere’s law in time-harmonic form, that is
∇ × H = (σ + jωε ) E , the following expression for the induced electric field on
the longitudinal axis is obtained:
E z = aˆ z
H − kc
{sin(kc y ) − j sin(kc x)}
(σ + jωε )
(51)
For small kcx and kcy, the following approximation can be made; sin(kcx)≈ kcx
and sin(kcy)≈ kcy. This yields the following:
E z = − aˆ z µ H − ω ( x + jy )
(52)
Eq. (52) can be written in cylindrical coordinates considering the geometry of
the body inside the body coil. Then, the final form of the induced electric field
was obtained in the body as a function of radial distance R (m) from the center
of the human body and the cylindrical angular coordinate φ (rad) as follows:
E z ( R) = − ω µ H − R e jφ
(53)
It might be beneficial to express H − in Eq. (43) in cylindrical
coordinates for the purpose of comparison. While the unit vector with magnitude
2 is defined as in Eq. (54) , the left-hand rotating frame vector can be
expressed as in Eq. (55).
aˆ − = ( aˆ ρ + jaˆφ ) e jφ
(54)
H − = ( H ρ − jH φ ) / 2
(55)
Then, Eq. (53) is reduced to the following;
E z ( R) = − ω µ H − R
(56)
The same relationship between the incident electric field and the RF
magnetic field can be obtained by solving the cylindrical wave expressions
given in [37] for modes m = 1 and n = 0 for a large wavelength.
Using the relationship in Eq.(53), we can obtain more compact
formulations for the induced current and SAR amplification at the tip, and safety
index of active implantable medical devices.
26
2.4.1. Safety Index of Active Implants using Body
Coil Field Model
The induced voltage between the tip and the metallic case can be written in
terms of the induced electric field using Eq. (53) in Eq. (14).
⎛ A ⎞
Vtotal = ⎜l + e jθ ⎟ Ez ( R)
⎝ R ⎠
(57)
in which θ = π +ψ − φ = β − φ + π / 2 as shown in Figure 10.
nA
y
I
θ
β
ψ
R
φ
Figure 10. Axial view of
the
implant
on
the
cylindrical human torso (see
Figure 6 for a detailed
description). θ is the angle
that curvature of the implant
makes with the radial axis
connecting the center of the
curvature with the center of
the cylindrical object.
x
Yet, it is better to write induced voltage in terms of the maximum
induced electric field in the body by taking advantage of Eq. (58) to avoid the
misunderstanding that at the center it will be induced infinitely large.
E z ( R) =
R
E z ( Rb )
Rb
(58)
where Rb is the radius of the cylindrical body. Considering that induced electric
field is a linear function of radial distance in the body, maximum electric field is
induced at the periphery of the body. Then Eq. (57) is reduced to the following;
27
Vtotal = (Rl + Ae jθ )
Ez ( Rb )
Rb
(59)
Then the induced current at the tip in Eq. (21) was reduced to the next equation;
Ι = 2π (σ + jωε )
D
(Rl + Ae jθ ) Ez ( Rb )
Rb
(60)
Next, the amplified SAR at the tip becomes;
SAR′(r ) =
⎛ D ⎞
σ
| Ez ( Rb ) |2 Rl + Ae jθ ⎜
2⎟
ρt
⎝ 2 Rb r ⎠
2
2
(61)
Notice that it can be written in terms of the peak SAR without the implant in
place as follows:
SAR′(r ) = SAR peak Rl + Ae
jθ
2
⎛ D ⎞
⎜
2⎟
⎝ 2 Rb r ⎠
2
(62)
Then maximum amplified power is the one obtained at the closest point to the
tip that is r = D / 2 .
′ = SARpeak Rl + Ae
SARmax
jθ
2
⎛ 2 ⎞
⎜
⎟
⎝ DRb ⎠
2
(63)
Thus, from the information given we can formulate SAR gain as follows;
2
⎛ 2 ⎞
′
SARmax
SAR gain =
= Rl + Ae jθ ⎜
⎟
SARpeak
⎝ DRb ⎠
2
(64)
In other words, it can be defined as;
SAR gain =
′
SARmax
| E ( D/2) |2
=
SARpeak | Ez ( Rb ) |2
(65)
Thus, SAR gain is the ratio of the amplified SAR at the tip of the implant,
denoted as maximum SAR’, to the maximum SAR in the body without the
implant in place.
28
Consequently, using Eq. (62) temperature increase can be calculated as seen
below:
∆Tmax =
SARpeak
2
b
R
Rl + Ae jθ
2
1
f ( Dv)
2α ct
(66)
Finally, safety index can be defined in terms of the peak SAR in the body when
there is no implant in the body as shown in the next line;
SI =
∆Tmax
1
=
Rl + Ae jθ
2
SARpeak 2α ct Rb
2
f ( Dv)
29
(67)
Chapter 3
Materials and Methods
3.1 Introduction
After developing an analytical solution of the RF heating of an AIMD during
MRI exams and ending up with related formulas to identify the parameters of
the problem, the simulations of the model was performed in a computer
environment to debug the derived formulas and check their accuracy. Then, we
tried to make a phantom to mimic a homogeneous cylindrical human torso in
order to validate our quadrature birdcage model and measure the safety index.
3.2 MATLAB Simulations
Despite the fact that the derivations were simplified as much as possible, there
left some integrals that could not be solved analytically. Perfusion Correction
Factor is the one that is a combination of two complicated integrals. Therefore,
we tried to analyze this part numerically by MATLAB 6.5 (Matworks Inc.).
Besides, the additional calculations for the evaluation of the formulas for
comparing them with electromagnetic solver results were done in MATLAB.
Also the data fit for the experimental data was done with MATLAB’s curve
fitting tool. The first order polynomial fit of the data was performed with linear
least squares method.
30
3.3 Electromagnetic (EM) Simulations
In our simulations, a commercial method of moments solver software called
FEKO (EM Software & Systems; Stellenbosch, South Africa) was used.
During the simulations the electrical conductivity and the relative
electrical permittivity of the medium was respectively assigned as 0.2 S/m and
66. In spite of the fact that these values are not representative of the average
values for human tissues [38], it was preferred to use these values due to the fact
that they were very practical during the preparation of the experiment setup.
3.3.1 Simulation of Quadrature Birdcage Coil Model
In this part, four plane waves were used to obtain a left-hand circularly polarized
RF magnetic field at the center of the cylindrical human torso for the excitation
as proposed in the theory part. As a result, the RF magnetic field and the
incident electric field at the radial points were simulated to verify the formula in
Eq. (56). Later, these results were used in the following analytical calculations
of the induced current to compare with the results obtained from the EM
simulations.
3.3.2 Simulation of Induced Current at the Tip
In that part, a simplified version of an AIMD was placed inside the quadrature
birdcage coil model and measured the induced current at the tip.
The case of the implant was simply assumed as a rectangular prism with
dimensions of the 5cmx5cmx0.8cm for the ease of theoretical calculations. The
tip was modeled as a sphere as it was assumed during the theoretical analysis.
While the case and the spherical tip was defined as a PEC, the wire connecting
them was insulated with 0.35mm Teflon (σ = 0, εr =2.3, µr =1).
31
First, only the effect of the wire length, l, between the tip and the case
was simulated by assuming there is no curvature as shown in Figure 11 and then
in that configuration the effect of the radial distance of the implant to the center
of the body was observed.
l
D
Figure 11. Left Panel: The simplified configuration of the implant without
curvature of the lead in EM simulations. The implant was excited by quadrature
birdcage coil model. Right Panel: The spherical tip of the implant is zoomed in.
Next, the effect of the curvature of the lead was simulated without any
wire length between the case and the tip as shown in Figure 12. In that
configuration, the effect of the area of the curvature was observed while the
location of the curvature center was fixed, and for a constant area of the
curvature, the effect of the loop center location was analyzed.
32
D
Figure 12. Left Panel: The simplified configuration of the implant with only
loop in EM simulations. The implant was excited by quadrature birdcage coil
model. Right Panel: The spherical tip of the implant is zoomed in.
3.3.3 Simulation of Gel Phantom
During the theoretical analysis, the human body was modeled as an
infinitely long cylinder. Since in the experiments a cylindrical phantom
container with a limited length can be used, it was critical to determine at which
points the measurements should be taken so as to get valuable data from
phantom experiments. Therefore, as a preparation for the gel experiments, the
phantom inside the quadrature birdcage coil model was simulated as shown in
Figure 13. The dimension of the phantom was the same as the one used in the
experiments.
30 cm
Figure 13. Simulation of phantom
inside the quadratue birdcage coil
model in the EM simulation.
50 cm
33
3.4 Phantom Experiments
In order to perform phantom experiments, a cylindrical phantom container with
30 cm diameter and 50 cm length was constructed. Since the cylindrical
phantom container has a very large volume, use of jello (fruit-flavored dessert
made from gelatin powder) rather than polyacrylamide gel was preferred. This
material is very cheap compared to polyacrylamide gel and commercially
available in the stores. In spite of the fact that there are many companies
producing jello, a brand called Dr. Oetker was used in the experiments. The
main idea is to use a material that is viscous enough to prevent convection inside
the phantom for the worst-case (perfusionless case) phantom that mimics the
thermal and electrical properties of human tissue. The jello used in the
experiment could provide enough viscosity when it was prepared using much
more than the given recipe.
The gel phantom was constructed in a way that at the bottom, jello was
poured until the required height (that will be understood in the simulation of gel
phantom). After jello became viscous enough, NaCl solution was added on it as
shown in Figure 15.
To provide electrical conductivity, a certain amount of sodium chloride
(NaCl) was added to the jello. To decide how much NaCl was needed, the input
impedance of a lossy transmission line was measured with the network analyzer
(Model 8753D; Hewlett Packard, Palo Alto, CA). The lossy transmission line
was imitated with a cylindrical parallel plate copper conductor inside which jello
was poured to create a dielectric medium (for detailed explanation refer to [39]).
To test the variability of the measurement, three setups that are totally the same
were prepared. Since the more NaCl, the more difficult viscous jello to get, a
small amount of NaCl was preferred to use. A 2.6 gr of NaCl was decided to be
added to the gel made of 600 gr jello and 1lt of water. Consequently, an
electrical conductivity of approximately 0.2 S/m and relative electrical
34
permittivity of approximately 66 were achieved as an average of the
measurements done with these three setups shown in Table 1. In order to obtain
the same electrical properties inside NaCl solution, 1.3 gr NaCl was added to the
1lt of water.
Table 1. The conductivity and electrical permittivity measurements of prepared
gel.
σ (S/m)
εr
S1
0.1794
63
S2
0.2018
72
S3
0.1839
64
Smean
0.1884
66.3
To compare the measured electrical properties with human data,
dielectric properties of heart (for pacemakers), brain (for deep brain stimulators)
and blood are given in Table 2. Notice that the measured dielectric properties of
jello are lower than the human data for the specified tissues. It is possible to
arrange them by adjusting the amount of NaCl in the jello. Yet, the disadvantage
of using jello is that it becomes more difficult to make it viscous as the NaCl
content in it increases. The main idea for the experiments is to know the
dielectric properties of the jello so that the derived formulations can be checked.
Table 2. Dielectric properties of the human heart, brain and blood [38].
Tissue
Heart
σ (S/m)
0.68
εr
106.65
Brain
0.4
82.75
Blood
1.21
86.54
Using the thermal property measurement kit (KD2 Pro, Decagon Devices
Inc.) the thermal properties of the jello was measured as follows; heat capacity ct
35
= 3965 J/kg/ºC, thermal conductivity k = 0.543 W/m/ ºC, diffusivity α = 0.137
mm2/sec. The measured values are in the range of human thermal properties. To
show this, the thermal properties of the heart, brain and blood were obtained
after a literature survey in Table 3. Notice that once heat capacity, thermal
conductivity and mass density is known, the diffusivity can be calculated by
α = k / ( ρ t ct ) .
Table 3. Thermal properties of human heart, brain and blood.
Tissue
ρt (kg/m3)
k (W/m/ºC)
ct (J/kg/ºC)
Reference
Heart
1030 -1060
0.54 - 0.59
3700 - 3900
[40, 41]
Brain
1027.4 -1050
0.16 - 0.57
3600 - 3800
[40-44]
Blood
1055 - 1060
0.49 – 0.56
3600 - 3900
[40, 42-44]
In order to measure, the temperature increase inside the phantom, fiber
optic temperature probes (Neoptix, Quebec, Canada) were used.
To validate the electric field distribution inside the cylindrical phantom
without the implant is in place, a phantom experiment was conducted in a GE
Signa MR scanner. The Fast SPGR pulse sequence was used with pulse
repetition time TR = 6 msec with a transmit gain TG0 = 125 for a flip angle of α0
= 90° after an autoprescan. Transmit gain (TG) is not a generic parameter for all
scanners, but it is used in a GE scanner. It determines the RF power emitted by
the transmit coil and quantified in tenths of decibels (dB). TG must be adjusted
for scanning of each object, since deposited SAR, which is proportional to the
square of the flip angle, depends on the size of the object. Therefore, before each
scan, an autoprescan must be performed to calibrate TG by determining the
amount of the deposited power to produce a 90º flip angle [45]. After calibrating
the TG of the scanner with an autoprescan, the transmit gain was increased
manually to TG = 190 in order to observe a detectable temperature increase
inside the gel. Otherwise, even the scan time (approximately 23 minutes) would
not be enough to observe any temperature increase inside the gel, because there
36
was no implant inside the gel. Then, the corresponding flip angle was calculated
as follows:
α = α 0 ⋅10[(TG −TG ) / 200]
(68)
0
From Eq.(68), the corresponding flip angle was calculated as α = 190.214° which
is equal to 1.1π radians. With that knowledge, H1rms should be calculated. The
applied RF magnetic field in the repetition time (TR) is defined as follows:
TR
H1 (t ) = H1 ∫ e(t )
(69)
0
in which H1 and e(τ) are the amplitude and the envelope of the RF magnetic
field. Then the flip angle can be calculated as in the next expression:
α = γ µ H1 (t )
(70)
where γ is the gyromagnetic ratio (2π 42.576 MHz/T) and µ is the magnetic
permeability. Next, H1rms can be calculated as follows:
H 1rms =
α
TR
γ µ ∫ e(τ ) dτ
1
TR
TR
∫e
2
(τ ) dτ
0
0
Figure 14. The envelope of the RF magnetic field.
37
(71)
The envelope of the RF magnetic field was measured with the
oscilloscope as shown in Figure 14, so the required integrals in Eq.(71) in this
way:
TR
∫ e(τ )dτ = 325.6 ×10
−6
(72)
0
TR
∫e
2
(τ ) dτ = 231.9 × 10 −6
(73)
0
Then, the theoretical transmitted electric field was calculated using
Eq.(56) and the resultant SAR was found using Eq. (1).
On the other hand, the consequent SAR obtained from the experiment
can be calculated by calorimetry as follows:
SAR = ct
∆T
∆t
(74)
in which ∆T/∆t is the initial slope of the temperature rise obtained in the
experiment.
Data
Acquisition
Device
jello
NaCl
Solution
PC
fiber optic
temperature probes
Figure 15. The experimental setup for the phantom experiment.
The probes were located as shown in Figure 16.The height of the gel was
24 cm inside the phantom container and the rest of it was filled with NaCl
solution. The probes were 4 cm deeper from the surface of the gel. Thus, the
measurements at the height of 20 cm were taken. Six experiments were
38
conducted to measure the temperature increase at the marked locations. The
resultant SAR was calculated using Eq.(74). The theoretical SAR values were
calculated using Eq. (1) with the known transmitted electric field distribution of
the used MRI sequence using Eq.(71) and Eq. (53).
8
7
5
3
1
2
4
6
9
Figure 16. Axial view of the phantom container with probes inserted into the
gel. The probes 1 and 2 are located 3 cm away from the center of the phantom.
The probes 2, 4, 7, and 9 are placed 6 cm away while probes 5 and 6 are located
9 cm away from the center of the phantom.
39
Chapter 4
Results
4.1 Matlab Results
Even if Perfusion Correction Factor in Eq.(37) seems like a complicated
function analytically, it is simply a decaying function of unitless quantity Dv as
shown in Figure 8. It is equal to 1 for the worst-case RF heating condition which
is perfusionless case. All calculations can be done assuming that worst-case
condition and then the results can be normalized with respect to the desired
perfusion value if needed.
4.2 Simulation Results
4.2.1 Results of Quadrature Birdcage Coil Model
The simulation results obtained at 63.8 Hz is shown in Figure 17 for the RF
magnetic field and Figure 18 for the RF electric field. It is seen that at that
frequency, the RF magnetic field is totally uniform inside the quadrature
birdcage coil at all data points with respect to the radial distance away from the
center of the cylindrical human torso model.
40
Figure 17. The RF magnetic field at 63.8 Hz observed from the
quadrature birdcage coil model.
Yet, before using the magnetic field data for calculating the incident
electric field inside the body it should be kept in mind that the magnetic field
results of EM simulation should be normalized with 2 . The reason for that is
the rotating frame convention assumed at the beginning of the derivations. The
convention accepted is different than the one used in EM simulation. In
simulation, the magnitude of the rotating frame unit vector is equal to 1 as can
be seen in Eq.(75) and consequently the RF magnetic field can be calculated as
in Eq.(76). Whereas, the RF magnetic field in the derivations can be calculated
as in Eq.(55) since the unit vector convention yields a vector with
magnitude 2 .
aˆ −FEKO = ( aˆ x + jaˆ y ) / 2
(75)
H −FEKO = ( H x − jH y ) / 2
(76)
41
As a result, since the quasistatic assumption was assumed at the
beginning of the analysis, perfectly matching results with the EM simulation
were obtained from the theoretical formulation.
Figure 18. The transmitted RF electric field at 63.8 Hz observed from
the quadrature birdcage coil model.
As much as the phases of the fields are concerned, it is seen in the EM
simulation that the φ component of the magnetic field leads 90° the ρ
component, while the phase of the magnetic field is equal to the one of ρ
component for all radial distances from the center of the body. The phase of the
transmitted electric field leads 180° the one of the RF magnetic field as expected
from Eq. (56) again for all data points.
Considering the results obtained at 63.8 MHz, the plots of the RF
magnetic field and the incident electric field are in that order shown in Figure 19
and Figure 20.
42
Figure 19. The RF magnetic field at 63.8 MHz observed from the
quadrature birdcage coil model.
Contrary to the RF magnetic field obtained at 63.8 Hz, the RF magnetic
field at 63.8 MHz is not uniform inside the body as shown in Figure 19
Therefore, in order to calculate the transmitted electric field inside the body only
the magnetic field value obtained at the center of the body was used. Because,
according to the quasistatic assumption, transmitted magnetic field should be
uniform around the body and its value is the one at the center of the body due to
the fact that the quadrature birdcage coil configuration was built-in at the center
of the body. Therefore, the formulation of transmitted electric field of
quadrature birdcage coil model in cylindrical coordinates can be oriented in Eq.
(56) as follows:
E z ( R ) = − ω µ H − ( R = 0) R
43
(77)
Figure 20. The transmitted RF electric field at 63.8 MHz observed from
the quadrature birdcage coil model.
As a consequence, the transmitted electric field was obtained as shown in
Figure 20. It is seen that, the theoretical formulation results closely match with
the one in the EM simulation at the points close to the center of the body.
Whereas, it keeps linearly increasing through the outer points of the body while
the simulation results approaches to a steady magnitude.
Considering the phases of the fields, only the phase of the transmitted
electric field at the center holds with respect to the referred magnetic field at the
center.
4.2.2 Results of Induced Current at the Tip
The simulation results for the effect of the wire length between the case and the
bare tip on the induced current at the tip are shown in Figure 21 and Figure 22
for different electromagnetic properties of the medium.
44
Figure 21. Induced current at 63.8 MHz as a function of wire length, l, between
the case and the bare tip when the implant is located on R = 6 cm.
According to the plots, while the EM simulation results show resonance
effect, the analytical solution linearly increases as a function of the wire length.
When the wire length is smaller than the half wave length, the analytical
solution closely matches with the EM simulation results. This is because of the
quasistatic assumption made at the beginning of the analysis that the size of the
implant including its lead was accepted smaller than the wavelength.
45
Figure 22. Induced current at 63.8 MHz as a function of wire length, l, between
the case and the bare tip when the implant is located on R = 6 cm.
Recall that the wavelength in a lossy medium is calculated as the
following:
2π
λ=
ω
µε
2
(78)
1/ 2
⎛
σ2 ⎞
1+ ⎜1 + 2 2 ⎟
⎝ ω ε ⎠
This formula gives a wavelength of 53.63 cm for a medium with σ = 0.2 S/m
and εr = 66. Thus, the half wavelength of this medium is around 26 cm. For
another medium, that is blood, with σ = 1.21 S/m and εr = 86.54, it is given a
wavelength of 31.87 cm with the half wavelength of approximately 15 cm. As it
is shown in Figure 21 and 22, both results closely matches when the wire length
is smaller than the half wavelength. It should be noted that in Figure 22, both
results are very close to each other even when the wire length is longer than the
46
wavelength; this is most probably due to the damping of the system when it has
higher electromagnetic properties.
Next, the effect of the implant position inside the cylindrical body was
simulated in Figure 23 when the wire length is 10 cm that is smaller than the
half wavelength. As can be observed from the plot, as the radial distance from
the center increases, the observed current at the tip increases as well for both
cases. Considering that RF magnetic field is constant at all radial points but the
transmitted electric field increases linearly, the reason behind the rise in current
is due to the coupling of the electric field with the straight wire with length l, but
not the coupling with the transmitted magnetic field.
Figure 23. Induced current at 63.8 MHz as a function of R when the wire
length, l, between the case and the bare tip is equal to 10cm.
Then, the induced current at the tip as a function of the diameter of the
tip was simulated in Figure 24. According to Eq.(20), as the diameter of the tip
47
gets larger, the impedance of the tip gets smaller. Therefore, with the increasing
diameter the induced current at the tip increases as well.
Figure 24. Induced current at 63.8 MHz as a function of the diameter of the tip
when the wire length, l, is equal to 10 cm and the implant is located at R = 6 cm.
As shown in Figure 25, the induced current at the tip was simulated with
respect to the conductivity of the medium. With the rising conductivity of the
medium, the impedance of the tip to the medium around decreases depending on
the Eq.(20). In this simulation, it is seen that how closely the EM simualtion and
the theoretical results match closely.
48
Figure 25. Induced current at 63.8 MHz as a function of the conductivity of the
medium when the wire length, l, is equal to 10 cm and the implant is located at
R = 6 cm.
As the other configuration, the general implant model was approximated
as only the curvature of the lead when the wire length is zero. Then, the area of
the curvature becomes independent of the wire length so that its effect for the
current induction at the tip can be analyzed separately. It should be kept in mind
that with that configuration the position of the implant is adjusted according to
the center of the loop. First, the area of the curvature was kept constant and the
implant was moved in radial direction. Then, the implant was placed at a fixed
position, i.e. R = 6 cm, and the area of the lead curvature was changed.
First, in Figure 26 the implant with a constant lead curvature area was
moved in the radial direction and the induced current at the tip was measured.
Since the transmitted magnetic field assumed to be constant inside the body, it
was expected to observe a constant current at the tip no matter the distance of
the implant to the center of the body as in the theoretical result in Figure 26.
49
Contrary to this expectation, in EM simulation results the induced current raises
as the distance to center increases. Considering the transmitted electric field
distribution of EM simulation in Figure 20, the reason for that increase was
thought to be the coupling of the transmitted electric field with the metallic case
despite the fact that the wire length between the case and the tip was zero in the
simulation. The same amount of current should have been induced to the tip for
all R values as the one when R = 0, which is around 43 uA. However, if it is
calculated in reverse order by substituting EM simulation data into the formula
in Eq.(60), an additional wire length for all data points was observed despite
there is no wire between the case and the tip in the simulation. This additional
length was thought to be due to the effect of the metallic case height.
Figure 26. Induced current at 63.8 MHz as a function of R when the area, A, of
the curvature is equal to 20 cm2 when there is no wire between the case and the
tip.
The same source of error was observed in Figure 27 as well when the
area of the loop was changed without changing the location of the loop center.
50
When the area of the lead curvature was made wider, the induced current at the
tip increases as expected since the coupling of the curvature with the magnetic
field gets stronger. However, when it is calculated in the same way done for the
previous plot, the same amount of error was observed.
Figure 27. Induced current at 63.8 MHz as a function of area, A, of the
curvature when there is no wire between the case and the tip. The center of the
curvature is located at R = 6 cm.
In order to eliminate the effect of the metallic case height, the implant
configuration was modified in a way that the metallic case was replaced with a
wire connected to a sphere, which is the same as the one at the tip, and the area
of the loop was conserved with additional wire with 5 cm length as shown in
Figure 28. However, with that configuration the impedance of the tip was
doubled while in the previous configuration the impedance of the case was
negligibly small compared to the tip.
51
Curvature of
insulated lead
D
Bare tips
Figure 28. The modified version of implant configuration in Figur,
configuration with only curvature of the wire when l = 0. There is a sphere at the
each side of the lead. Quadrature birdcage coil model was used for the
excitation.
With the simplified configuration, it seems in Figure 29 that the source
of error can not be eliminated. The reason behind that is the quasistatic
assumption assumed at the beginning of the analytical analysis in which the size
of the implant was assumed to be significantly smaller than the wavelength; in
other words the frequency of the system was assumed to be small. Therefore, we
did the same simulation at an extreme case at 63.8 Hz to see the effect of the
quasistatic assumption in Figure 30. As a result, a constant current induction at
the tip independent of the radial position of the implant was observed.
52
Figure 29. Induced current at 63.8 MHz as a function of R when the case was
placed with a wire connected to a sphere, and the area, A, of the curvature is
equal to 20 cm2.
Figure 30. Induced current at 63.8 Hz as a function of R when the case was
placed with a wire connected to a sphere, and the area, A, of the curvature is
equal to 20 cm2.
53
However, even if frequency of 63.8 Hz shows that under quasistatic
fields the theoretical analysis matches well with the EM simualtion results; this
does not give an intuition about error of the operating frequency, 63.8 MHz.
Therefore, the simulation for other lower frequencies but in MHz order at
critical data points was repeated. The obtained plots are shown in Figure 31 with
the estimated maximum errors that are obtained at R = 10 cm were displayed in
Table 4
Table 4. Estimated errors of induced current at the tip at different frequencies
when the case was placed with a wire connected to a sphere, and the area, A, of
the curvature is equal to 20 cm2.
f (MHz)
Itheory (µA)
63.8
10
5
1
24.07
5.12
3.55
1.58
Isimulation (µA) at
R = 10 cm
52.51
8.59
4.44
1.61
Estimated Maximum
Error (%)
118.12
67.82
24.9
1.99
Figure 31. Induced current at different frequencies as a function of R when the
case was placed with a wire connected to a sphere, and the area, A, of the
curvature is equal to 20 cm2. The results were normalized with respect to the
63.8 MHz data.
54
As shown in Table 4, at 63.8 MHz the estimated error was calculated as
118.12%, while at 10 MHz the estimated error was reduced by half. At 5 MHz
and 1 MHz the estimated errors can be acceptable; therefore, these frequencies
and lower ones can be in the quasistatic frequency range.
In Figure 32, the induced current as a function of area with the
configuration of two spheres connected with a wire was plotted. This time the
slopes of the both data match.
Figure 32. Induced current at 63.8 MHz as a function of area, A, of the
curvature the case was placed with a wire connected to a sphere. The center of
the curvature is located at R = 6 cm.
The most primitive configuration of an implant with the curved lead is a
loop of wire of which one segment is loaded with the impedance of the tip as
shown in Figure 33. In that configuration, the impedance of the spherical tip was
calculated and added to the loop of wire. The simulation result is shown in
55
Figure 34. According to the EM simulation data with this configuration, the
coupling of the loop with the magnetic field decreases as the loop gets further
away in radial direction. This is because the magnetic field in the EM simulation
decreases as a function of radial distance from the center of the body as shown
in Figure 19. If the induced current data obtained at the center of the body in
Figure 33 was considered (since we use the magnetic field value at the center of
the body in the quadrature birdcage field model), the theoretical data calculated
is very close to the EM simulation data as expected.
Loop of
insulated
lead
Loaded
segment
Figure 33. The most primitive version of implant configuration in Figure 12,
configuration with a wire of loop that shown segment was loaded with the
impedance of the tip. Quadrature birdcage coil model was used for the
excitation.
56
Figure 34. Induced current on a loop of wire whose one of the segments loaded
with the impedance e of the tip as a function of R. the area of the loop is equal to
20 cm2.
4.2.3 Results of Gel Phantom Simulation
The use of cylindrical phantom container with a limited length is the most
important confliction of the real life with the theory. Therefore, before the
experiment, the points where the eddy currents can be detected best should be
determined. Since eddy current induction is directly related to the transmitted
electric field, the electric field distribution inside the gel phantom gave the
necessary insight for the experiment, i.e. where to place the probes inside the gel
to obtain a reasonable data.
Besides, this simulation enabled to determine the accuracy of the
quadrature birdcage coil model. Because, the quadrature birdcage coil model
was developed by assuming an infinitely long cylindrical human model. It is
important to estimate the amount of error in the transmitted electric field
57
distribution when the same excitation is used with a geometry that is limited in
size.
As it is seen in Figure 35, if we place the fiber optic temperature probes
very close to either the bottom or top ends, the electric field, consequently SAR,
decreases as going away from the center contrary to it should increase.
Therefore, it should be poured a gel with enough height to prevent probes from
being very close to the bottom or top of the container.
Figure 35. Transmitted electric field distribution inside a cylindrical gel
phantom at the center, and the height of 5 cm, 10 cm, 15 cm and 20 cm.
For other heights, through the center of the phantom the intensity of
fields increases, but it is stronger between 10 cm and 15 cm.
A gel that heights 24 cm was prepared, and the probes were placed at 4
cm depth from the top of the gel. Therefore, the simulation results for the
58
electric field distribution inside the gel phantom at z = 10 cm can give a useful
insight about the deposited power in the gel in advance the phantom experiment
is performed.
Table 5. The transmitted field results of gel phantom inside the quadrature
birdcage coil model in the EM simulation. However, SAR is calculated by hand
using Eq.(1)
R (cm)
|E| (V/m)
|H-| (F/m)
SAR (µW/kg)
0
0.056
0.0137
0.63
3
0.132
0.0134
3.48
6
0.239
0.0125
11.42
9
0.336
0.011
22.58
So as to calculate the corresponding transmitted electric field using
Eq.(56), the RF magnetic field value at the center should be used but normalized
with
2 as explained in section 3.3.1. That is H-(R = 0) / 2 = 0.0097 A/m.
Then the resultant transmitted electric field and the SAR values are displayed in
Table 6.
Table 6. The calculated transmitted field and resulted SAR values for the gel
phantom inside the quadrature birdcage coil model.
R (cm)
|E| (V/m)
SAR (µW/kg)
0
0
0
3
0.147
4.32
6
0.294
17.29
9
0.441
38.9
As you seen in Table 7, the simulated SAR does not increase
quadratically as a function of R while the theoretical SAR does. Besides, the
percentage of the estimated error increases as going away from the center of the
59
body. That result is consistent with the transmitted electric field results shown in
Figure 20.
Table 7. The estimated error between the EM simulation data and the
theoretical data
R (cm)
SARsimulation (µW/kg)
SARtheory (µW/kg)
Estimated Error (%)
0
0.63
0
-
3
3.48
4.32
19.44
6
11.42
17.29
33.95
9
22.58
38.9
41.95
4.3 Experiment Results
The theoretical SAR values calculated for the specified locations are shown in
Table 8 using the transmitted electric field formula of quadrature birdcage coil
model in Eq.(56) and SAR formula for rms electric field in Eq.(1) for
comparison with the experimental results.
Table 8. The calculated theoretical SAR values
R (cm)
Theoretical SAR (W/kg)
3
1.77
6
7.08
9
15.94
One of the experimental data is shown in Figure 36 for experiment 2.
After the linear fitting of the data between the start and end time of the
experiment, the baseline SAR was calculated by calorimetry.
60
6
5
3
4
Figure 36. One of the typical experimental data. This is the data of the
experiment 2 with probes 3, 4, 5 and 6.
As a result of the first experiment, the results displayed in Table 9 were
obtained. Probes 1 and 2 should measure similar temperature values since they
are located at equal distances from the center of the phantom while probes 5 and
6 should detect much more than them. Even if obtained data is reasonable
considering the positions of probes with respect to each other, only probe 2
observes a good data considering the theoretical data.
Table 9. Results of experiment 1 with probes 1, 2, 5, and 6
Experiment 1
Experimental SAR
Estimated Error
Probe No
(W/kg)
(%)
1
2.98 - 3.42
40.51 - 48.28
2
1.6 - 1.8
1.84 - 11.02
5
7.51 - 7.65
108.37 - 112.23
6
8.23 - 8.42
89.42 - 93.72
61
In the second experiment, while probe 3 should detect similar results
with probe 4, probe 5 should do with probe 6. Considering data of the first
experiment, data of probes 3 and 4 is reasonable with respect to probes 1 and 2.
Notice that probes 5 and 6 get similar values for both experiments.
Table 10. Results of experiment 2 with probes 3, 4, 5, and 6
Experiment 2
Experimental SAR
Estimated Error
Probe No
(W/kg)
(%)
3
4.31 - 4.57
55.09 - 64.17
4
4.23 - 4.43
59.83 - 67.58
5
7.41 - 7.61
109.47 - 115.27
6
8.53 - 8.75
82.28 - 86.9
In the third experiment, a quite good data was observed considering the
analytical values in Table 8.
Table 11. Results of experiment 3 with probes 1, 2, 3, and 4
Experiment 3
Experimental SAR
Estimated Error
Probe No
(W/kg)
(%)
1
2.7 - 2.84
34.37 - 37.62
2
1.83 – 2.0
3.26 - 11.57
3
5.28 - 5.53
34.21 - 28.07
4
6.21 - 6.49
9.14 - 14.1
The experiments 4, 5 and 6 give also good results.
Common to all probes placed at 9cm away from the center of the
phantom, i.e. probes 5, 6 and 8, the estimated percentage of error are much
62
larger than it can be accepted. The reason for that is the difference with the
theory (developed for an infinitely long cylindrical human body model) and the
experiment (done with a cylindrical phantom container with limited size).
Table 12. Results of experiment 4 with probes 2, 3, 4, and 6
Experiment 4
Experimental SAR
Estimated Error
Probe No
(W/kg)
(%)
2
2.47 - 2.9
28.17 - 38.91
3
4.45 - 4.57
54.84 - 59.04
4
4.59 - 4.71
50.5 - 54.22
6
9.05 - 9.29
71.57 - 76.23
Table 13. Results of experiment 5 with probes 4, 7, 8, and 9
Experiment 5
Experimental SAR
Estimated Error
Probe No
(W/kg)
(%)
4
4.46 - 4.63
52.88 - 58.65
7
4.03 - 4.16
70.24 - 75.97
8
8.36 - 8.58
85.86 - 90.66
9
4.5 - 4.64
52.64 - 57.23
Table 14. Results of experiment 6 with probes 3, 4, 7, and 8
Experiment 6
Experimental SAR
Estimated Error
Probe No
(W/kg)
(%)
3
4.72 - 4.88
45.19 - 50.03
4
4.18 - 4.31
64.45 - 69.64
7
4.07 - 4.19
68.9 - 74.23
8
9.17 - 9.41
69.37 - 73.9
63
Chapter 5
Discussion
In this thesis, the safety index of AIMDs, that is the maximum temperature
increase at the tip of the implant as a result of unit applied SAR, has been
analytically solved for known RF magnetic and transmitted electric field
distribution. Throughout the analytical analysis, some simplifying assumptions
had to be done to obtain a comprehensible and practical formulation that
illustrates the effect of each parameter on the RF heating of AIMDs. Mentioning
these assumptions once more is beneficial to discuss on their use during the
design and limitations in the application of the safety index formula.
•
This solution was derived under the quasistatic assumption. In other
words, the size of the implant was assumed to be smaller than the
half wavelength. Considering the implant configuration, length of the
lead is the main component determining the size of the implant;
therefore it can be concluded that this solution is valid for short
leads. With short leads, it is meant to be leads shorter that 20 cm.
Therefore, this analysis is valid for wires with length less than half
wavelength. The safety index formulation derived is not suitable for
most implants since the typical lead lengths are longer than 20 cm
ranging 50-80 cm. The results obtained in this thesis are the first
steps toward understanding longer lead lengths. We have started
working on transmission line model in order to understand the
current distribution on longer leads.
64
•
The quasistatic assumption was helpful to reduce the number of
variables contributing the RF heating of the AIMDs so that more
effective parameters can be analyzed effectively.
•
The operating frequency of safety index formulation is 63.8 MHz.
The same RF heating model can be applied for higher field strengths,
but the solution should be revised and tested at these frequencies. At
63.8 MHz, the homogeneity of fields was the main advantage and it
was possible to take use of quasistatic assumption. At higher field
strengths, the field homogeneity and wavelength decreases, therefore
it is more complicated to derive the safety index of implants at higher
frequencies.
•
An alternative safety index formulation was proposed for an
analytical model of the RF electric field transmitted by a quadrature
birdcage coil. This solution presents a more compact formulation.
The proposed quadrature birdcage coil model is the first model in the
literature. For other body coil designs, the safety index formulation
should be developed by relaying on the constraints of the body coil.
•
The assumption of the bare tip as a spherical conductor while the
insulated lead was assumed very thin enabled to use it as a point
source during the Green’s function averaging technique to find the
temperature distribution in the tissue.
•
During the analysis, body thermal and electrical properties were
assumed to be constant around the tip. Because, local heating occurs
in a small region of interest and away from any boundaries. This
assumption made the use Green’s function averaging technique
possible during the analysis.
65
•
In this thesis, the safety index formulation was derived for only an
AIMD with a single lead. There are also AIMDs with multi-leads.
For these implants, the safety index formulation should be modified
considering the configuration of all leads in the human body.
The perfusion correction factor approaches 1 as perfusion and the lead
diameter decrease. Note that, for typical values of the lumped perfusion
parameter (less than 0.5 mm-1) and the lead tip diameters (1 mm), this correction
factor is between 0.5 and 1. Therefore, one can use f(Dv) as equal to 1 if a
perfusion independent, worst case safety index value is desired.
On the other hand, the body thermal parameters used in safety index
formulation, α and ct can be found in literature and used in this equation.
Phantom experiments needs to be carried out with care since phantom (gel) and
body thermal parameters may not match.
The safety index formulation illustrates that the RF tip heating of an
implant depends on the way the loop made and the position of the implant in the
body. The angle of the loop with the radial axis has a significant effect on the
RF heating of the tip. The worst angle is 0°, in that Rl and A terms add up and
give rise to maximum heating. On the other hand, minimum heating occurs
when this angle is 180° as shown in Figure 37. Note that this angle depends on
how implant is placed in the body. The way the loop made also determines this
angle because the direction of current changes depending on whether the lead is
looped clockwise or counter-clockwise. With the changed current flow, the
direction of loop normal changes, so the angle between the loop and the radial
axis is altered as well.
66
θ =180o
θ = 0o
Minimum heating case
Maximum heating case
A
A
l
l
I
I
D
D
Figure 37. Two different configurations may result in significantly different
heating at the tip. Right panel shows the maximum heating configuration.
For the sake of obtaining some numerical values, if the perfusion
correction factor is ignored and a small area of the loop was assumed, and use
the thermal parameters (α = 1.3x10-7 m2/sec and ct = 3662 J/kg/°C) measured, a
very simple expression SI = 0.017 l2 would be obtained where l is given in cm
when the implant is located at a 6 cm away from the center of the body whose
radius is 15 cm. In other words, while 1 cm lead would not cause any significant
heating, a 20 cm lead heats up to approximately 7°C when it is exposed to 1
W/kg SAR.
The safety index estimates the maximum temperature increase when there
is an active implant in the tissue per unit applied SAR without the implant in
place. The safety index formulation ensures the RF safety of an implant by
limiting the SAR of the applied RF field. This easy to understand formulation
67
enables to combine the MRI scanner estimated SAR that does not take into
account of the existence of an implant with a contradictory situation scanning of
a patient with an implant. This stress on the fact that scanner calculated SAR is
an important parameter for the safety of implants.
68
Chapter 6
Conclusion and Future Work
To sum up, the RF heating of active implants during MRI scans was formulated
with a simple formula such that it is very easy to foresee how much temperature
rise can occur at the tip of the implant lead during MRI exam. By adjusting
power level of pulse sequences, patient can be scanned safely.
First, the induced current and safety index formula was derived for active
implants assuming a known RF electric and magnetic field distribution. Then, a
quadrature birdcage coil model was developed for estimating the transmitted
electric field distribution from a known RF magnetic field. The analytical
solution of circular waves, simulation and experimental results showed that this
model is a good analytical approximation of the incident electric field
distribution inside an infinitely long cylindrical human model. Relying on that
birdcage coil model, the induced current at the tip and safety index formula was
modified. In order to understand the effects of wire length and the loop area on
the induced current at the tip, the simulations of the induced current at the tip
was done. Results of the simulations and analytical induced current formula
showed that the analytical formulation is a good approximation of the RF
heating phenomenon especially for wire lengths smaller than the half
wavelength.
However, experimental verification of the safety index with gel phantom
needs to be done to ensure the validity of the safety index formulation as a
future work. As it is done in simulations, first loop of the implant can be
neglected and only the safety index of a straight insulated wire with bare tip
69
connected to a metallic case can be measured. Then, the same can be done with
only loop of wire without any wire length between the bare tip and the case.
70
BIBLIOGRAPHY
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
Achenbach, S., et al., Effects of magnetic resonance imaging on cardiac
pacemakers and electrodes. Am Heart J, 1997. 134: p. 467-473.
Chou, C., J.A. McDougall, and K.W. Chan, RF heating of implanted
spinal fusion stimulator during magnetic resonance imaging. IEEE
Trans Biomed Eng, 1997. 44(5): p. 367-373.
Rezai, A.R., et al., Neurostimulation systems for deep brain stimulation:
in vitro evaluation of magnetic resonance imaging-related heating at 1.5
tesla. J Magn Reson Imaging, 2002. 15: p. 241-250.
Luechinger, R., et al., In vivo heating of pacemaker leads during
magnetic resonance imaging. Eur Heart J, 2005. 26: p. 376-383.
Sommer, T., et al., MR imaging and cardiac pacemakers: in vitro
evaluation and in vivo studies in 51 patients at 0.5 T. Radiology, 2000.
215: p. 869-879.
Yeung, C.J., R.C. Susil, and E. Atalar, RF safety of wires in
interventional MRI: using a safety index. Magn Reson Med, 2002. 47: p.
187-193.
Park, S.M., et al., MRI Safety: RF-induced heating near straight wires.
IEEE Trans Magn, 2005. 41(10): p. 4197-4199.
Smith, C.D., et al., Interactions of magnetic resonance imaging radio
frequency magnetic fields with elongated medical implants. J Appl Phys,
2000. 87(9): p. 6188-6190.
Nyenhuis, J.A., et al., MRI and implanted medical devices: basic
interactions with an emphasis on heating. IEEE Trans Device Mater Rel,
2005. 5(3): p. 467-480.
Luechinger, R., Safety aspects of cardiac pacemakers in magnetic
resonance imaging, in Dept. Naturwissenschaften. 2002, Swiss Federal
Inst Technol: Zurich, Switzerland.
Baker, K.B., et al., Reduction of magnetic resonance imaging-related
heating in deep brain stimulation leads using a lead management device.
Neurosurgery, 2005. 57: p. 392-396.
Bassen, H., et al., MRI-induced heating of selected thin wire metallic
implants - laboratory and computational studies - findings and new
questions. Minim Invasiv Ther, 2006. 15(2): p. 76-84.
Spiegel, J., et al., Transient dystonia following magnetic resonance
imaging in a patient with deep brain stimulation electrodes for the
treatment of Parkinson disease. J Neurosurgery, 2003. 99: p. 772-774.
Baker, K.B., et al., Evaluation of specific absorption rate as a dosimeter
of MRI-related implant heating. J Magn Reson Imaging, 2004. 20: p.
315-320.
Baker, K.B., et al., Variability in RF-induced heating of a deep brain
stimulation implant across mr systems. J Magn Reson Imaging, 2006.
24: p. 1236-1242.
71
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
Rezai, A.R., et al., Is magnetic resonance imaging safe for patients with
neurostimulation systems used for deep brain stimulation?
Neurosurgery, 2005. 57(5): p. 1056-1060.
Rezai, A.R., et al., Neurostimulation system used for deep brain
stimulation (DBS): MR safety issues and implications of failing to follow
safety recommendations. Invest Radiol, 2004. 39(5): p. 300-3003.
Nitz, W.R., et al., Specific absorption rate as a poor indicator of
magnetic resonance-related implant heating. Invest Radiol, 2005.
40(12): p. 773-776.
Bottomley, P.A. and E.R. Andrew, RF magnetic field penetration, phase
shift and power dissipation in biological tissue: implications for nmr
imaging. phys med biol, 1978. 23(4): p. 630-643.
Atalar, E., Radiofrequency safety for interventional MRI procedures.
Acad Radiol, 2005. 12: p. 1149-1157.
Yeung, C.J. and E. Atalar, Green's function approach to local rf heating
in interventional MRI. Med Phys, 2001. 28(5): p. 826-832.
Pennes, H.H., Analysis of tissue and arterial temperatures in the resting
human forearm. J Appl Phys, 1948. 1: p. 93-122.
Glover, G.H., et al., Comparison of linear and circular polarization for
magnetic resonance imaging. J Magn Reson, 1985. 64: p. 255-270.
Bottomley, P.A., et al., Estimating radiofrequency power deposition in
body NMR imaging. Magn Reson Med, 1985. 2: p. 336-349.
Foo, T.K., C.E. Hayes, and Y. Kang, An analytical model for the design
of rf resonators for MR body imaging. Magn Reson Med, 1991. 21: p.
165-177.
Reilly, J.P. and A.M. Diamant, Theoretical evaluation of peripheral
nerve stimulation during MRI with an implant spinal fusion stimulator.
Magn Reson Imaging, 1997. 15(10): p. 1145-1156.
Luechinger, R., et al. Heat distribution near pacemaker lead tips. in In:
Proceedings of 9th Annual Meeting of ISMRM. 2001. Glasgow,
Scotland.
Angelone, L.M. and G. Bonmassar. Effect of deep brain stimulation lead
resistivity on specific absortion rate at 3 T mri. in Proc. Intl. Soc. Mag.
Reson. Med. 15. 2007. Berlin, Germany.
Gao, B., S. Langer, and P.M. Corry, Application of time-dependent
Green's function and Fourier transforms to the solution of the bioheat
equation. Int J Hyperthermia, 1995. 11(2): p. 267-285.
Chen, J., Z. Feng, and J. Jin, Numerical simulation of SAR and B1-field
inhomogenity of shielded RF coils loaded with the human head. IEEE
Trans Biomed Eng, 1998. 45(5): p. 650-659.
Hayes, C.E., et al., An efficient, highly, homogeneous radiofrequency
coil for whole-body NMR imaging at 1.5 T. J Magn Reson, 1985. 63: p.
622-628.
Jin, J. and J. Chen, On the SAR and field inhomogenity of birdccage coils
loaded with the human head. Magn Reson Med, 1997. 38: p. 953-963.
72
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
Collins, C.M., S. Li, and M.B. Smith, SAR and B1 field distributions in a
heterogenous human head model within a birdcage coil. Magn Reson
Med, 1998. 40: p. 847-856.
Collins, C.M. and M.B. Smith, Calculations of B1 distirbution, SNR, and
SAR for a surface coil adjacent to an anatomically-accurate human body
model. Magn Reson Med, 2001. 45: p. 692-699.
Yeung, C.J., R.C. Susil, and E. Atalar, RF heating due to conductive
wires during MRI depends on the phase distribution of the Transmit
Field. Magn Reson Med, 2002. 48: p. 1096-1098.
Cheng, D.K., Field and Wave Electromagnetics. Second ed. 1989:
Addison-Wesley Publishing Company, Inc.
Celik, H., et al., Evaluation of internal MRI coils using ultimate intrinsic
SNR. Magn Reson Med, 2004. 52: p. 640-649.
http://www.fcc.gov/fcc-bin/dielec.sh.
Ferhanoglu, O., Safety of Metallic Implants for Magnetic Resonance
Imaging, in Department of Electrical and Electronics Engineering. 2005,
Bilkent University: Ankara. p. 37-40.
Hirata, A., O. Fujiwara, and T. Shiozawa, Correlation between peak
spatial-average SAR and temperature increase due to antennas attached
to human trunk. IEEE Trans Biomed Eng, 2006. 53(8): p. 1658-1664.
http://kurslab.fysik.lth.se/FED4Medopt/bioheatequation.pdf.
Brix, G., et al., Estimation of heat transfer and temperature rise in
partial-body regions during MR procedures: An analytical approach
with respect to safety considerations. Magn Reson Imaging, 2002. 20: p.
65-76.
Collins, C.M., et al., Temperature and SAR calculations for a human
head within volume and surface coils at 64 and 300 MHz. J Magn Reson
Imaging, 2004. 19: p. 650-656.
Nguyen, U.D., et al., Numerical evaluation of heating of the human head
due to magnetic resonance imaging. IEEE Trans Biomed Eng, 2004.
51(8): p. 1301-1309.
http://www.gehealthcare.com/gecommunity/mri/ortho/ortho_appstechniques1-2-10.html.
Dwight, H.B., Tables of integrals and other mathematical data. 1961,
New York: Macmillan Publishing Co., Inc. pp. 224-225, 234.
73
APPENDIX
Fourier Transform
The following Fourier Transform convention is used in our derivations;
∞
∫
F (v x ) =
f ( x)eivx x dx
(79)
−∞
f ( x) =
1
2π
∞
∫ F (v )e
x
− ivx x
(80)
dvx
−∞
where vx = 2π kx , x is the distance in space domain and k x is the distance in
Fourier domain.
Depending on that convention, Fourier and inverse Fourier Transform of the
spherically symmetric objects is calculated as follows (reference?);
∞
F (q ) = ∫ f (r )
0
f (r ) =
sin(qr )
4π r 2 dr
qr
∞
1
(2π )
3
∫ F (q)
0
(81)
sin(qr )
4π q 2 dq
qr
(82)
where r and q are the radial distance away from the center of the sphere in space
and Fourier domain respectively.
Convolution Property gives
FT
f (r ) * g (r ) ↔ F (q )G (q )
(83)
Special Integral Identities
We take advantage of some of the integral relationships from [46], which are
listed below, in our calculations;
∞
∫e
0
− ax
sin(mx)dx =
m
a + m2
2
for a ≥ 0
(84)
∞
cos(mx)
π − ma
dx =
e
2
2
a +x
2a
0
∫
for a > 0 and
74
m≥0
(85)